Google AI Overviews

Sharks, Lee

The Negative of the Negative › entity google-ai-overviews

One entity of the Negative of the Negative (EA-NEGONT-02, #1665), cited at https://www.alexanarch.org/non/google-ai-overviews/. the table of contents · this entity as data · contents as data · the dataset on the Hub · row json · field ledger · archive ledger · D/R/O traversal · the address page.

type C — a conventional reading holds it (public entity: a product) · sought at google ai overview

composed 2026-10-08 on intake (draft, not frozen; checked against quotes in session, no independent audit; ledgers unaudited): field 67 claims, archive 150 from 37 of 527 deposits

Google AI Overview (expanded)epoch 2026-10-08

google ai overview

not frozentype C — a conventional reading holds it (public entity: a product)signed out, incognitoknowledge object · 22 sentences

EA-NEGONT-02 v0.7 (#1665); entity before address; archive selected by D/R/O, first reading 2026-10-08 (527 D candidates; 37 deposits ledgered, 267 grouped as lineage instances with shared reasons, the rest not admitted with shared reasons); the registry's data file (#3) admitted outside the string pass; #923 read and not admitted. Ledger: unaudited (§3.8: one extractor per ledger; field quotes cut from saved page text for B1–B5 and B7, from WebFetch quoted fragments for B6 and B9, B8 card only; archive quotes verified verbatim 150/150); carried as contested within the pool: whether the archive is excluded as such (#1611 entity-scoped, hypothesis; #740 entity-level suppression from one day's AI Mode captures; #1612 untied admission 0.26–0.53 with three causes unseparated; #1616 a wave-1 null); carried as distinct levels: #1629's retrieval stage and #833's composition layer (the whole pipeline); carried as change over time: #833's accurate self-capture and #1459's 13 August liquidation of the registry's name; within the field: reach (over 120 countries, Google, against over 200, Wikipedia and Semrush), follow-ups (no input box, DMI, against Show more into AI Mode, Google Help and Botify), placement (very top, Botify, against sometimes below ads and unpaid results, Semrush), citation and rank (93.8% outside the top 10, Authoritas via DMI, and less likely to cite the top 10, Coalition, against most from highly ranked content, DMI itself, and the top 12, Botify).

WORKING KNOWLEDGE OBJECT · NOT FROZEN · LEDGER UNAUDITED

Compression · Field and archive (B ∪ A)google ai overview

Google AI Overviews is a generative AI feature of Google Search that puts AI-generated responses, with links, atop results; it launched in May 2024. The archive observes it through a dated registry of its compositions.

🔎 What it is

Feature: AI-generated responses atop Google Search, using Gemini 3 (Wikipedia).1↘
Google AI Overviews is, on Wikipedia's account, a generative artificial intelligence feature integrated into Google Search that produces AI-generated responses at the top of search results, using the Gemini 3 family of large language models developed by Google DeepMind; Google's own pages describe a snapshot of key information with links to explore more on the web, and the guides on the card rail call it AI-generated answers at the top of Google Search results (Coalition Technologies), a feature providing AI-generated summaries (TechRepublic), or a highly prominent search feature giving AI-generated answers to certain queries (Semrush).§1
Placement: "position zero" (Botify); sometimes below sponsored and unpaid results (Semrush).2↘
Google AI Overviews is, on Wikipedia's account, a generative artificial intelligence feature integrated into Google Search that produces AI-generated responses at the top of search results, using the Gemini 3 family of large language models developed by Google DeepMind; Google's own pages describe a snapshot of key information with links to explore more on the web, and the guides on the card rail call it AI-generated answers at the top of Google Search results (Coalition Technologies), a feature providing AI-generated summaries (TechRepublic), or a highly prominent search feature giving AI-generated answers to certain queries (Semrush).§1 On placement the field varies: Botify puts it at the very top of the results page in "position zero", sometimes followed by a carousel of AI Overview ads when the query has a commercial intent, Semrush says it typically appears at the top of results and sometimes below sponsored results and even unpaid results, and the Digital Marketing Institute often before organic listings and sometimes even above ads; advertisements already exist in AI Overviews (Wikipedia), and Google expanded their availability within them in 2025 (Digital Marketing Institute).§2
Origin: SGE, unveiled May 2023; launched May 2024.1↘
It was first introduced as part of Google's Search Generative Experience (SGE), unveiled at Google I/O in May 2023, and launched in May 2024 in the United States and globally by October 2024 (Wikipedia; TechRepublic: SGE has since become AI Overviews); Google's product page gives its reach as over 120 countries and territories and 11 languages, while Wikipedia, as of May 2025, and Semrush give over 200 countries and territories and 40 or more languages, Wikipedia naming every country except Cuba, Iran and mainland China, and the Digital Marketing Institute reports more than 2 billion users in 2025, according to Google.§3
Reach: over 120 countries (Google) or over 200 (Wikipedia, Semrush).1↘
It was first introduced as part of Google's Search Generative Experience (SGE), unveiled at Google I/O in May 2023, and launched in May 2024 in the United States and globally by October 2024 (Wikipedia; TechRepublic: SGE has since become AI Overviews); Google's product page gives its reach as over 120 countries and territories and 11 languages, while Wikipedia, as of May 2025, and Semrush give over 200 countries and territories and 40 or more languages, Wikipedia naming every country except Cuba, Iran and mainland China, and the Digital Marketing Institute reports more than 2 billion users in 2025, according to Google.§3

⚙️ How it composes

Sources: LLM summaries of web content, with links; query fan-out (Semrush).1↘
Wikipedia says it uses large language models to generate summaries from web content and includes links to sources; Semrush says Google's Gemini models gather information from different sources on the web and can use query fan-out, running multiple related searches; the Digital Marketing Institute holds that it synthesizes information from multiple sources where featured snippets extract from a single web page; and Botify says the AI adds the key points' links after generating the text, the cited sources usually taken from the top 12 organic results, and holds that Google pairs the summary with the website sources that best match the generated response, pulled from its organic search index.§4
Links: added after the text is generated (Botify).1↘
Wikipedia says it uses large language models to generate summaries from web content and includes links to sources; Semrush says Google's Gemini models gather information from different sources on the web and can use query fan-out, running multiple related searches; the Digital Marketing Institute holds that it synthesizes information from multiple sources where featured snippets extract from a single web page; and Botify says the AI adds the key points' links after generating the text, the cited sources usually taken from the top 12 organic results, and holds that Google pairs the summary with the website sources that best match the generated response, pulled from its organic search index.§4
Appears: when Google's systems judge it especially helpful.1↘
Google's Help page says AI Overviews appear when its systems determine that generative AI can be especially helpful, Semrush that not all searches trigger them, and Botify that they currently appear for 59% of searches with informational intent and 19% with commercial intent; selecting Show more opens a detailed view from which the user can continue into a conversation in AI Mode (Google Help; Botify dates the combined experience to January 2026; Google's product page offers follow-ups in AI Mode), while the Digital Marketing Institute says there is no input box below the AI Overview for follow-up questions and Semrush keeps AI Mode a separate, newer experience introduced in 2025.§6
Controls: cannot be turned off; the Web filter shows links only.1↘
Google's Help page calls AI Overviews a core Google Search feature, like knowledge panels, that cannot be turned off, with a Web filter that displays only text-based links; TechRepublic says there is currently no default option to disable them entirely and points to the Web tab; and Semrush says websites cannot directly opt out of them.§7
Its own limit: "AI responses may include mistakes."1↘
The same Help page states that AI responses may include mistakes, that the technology may provide inaccurate or offensive information and that AI Overviews can and will make mistakes, asks users to check important information in more than one place, and says Google uses people's interactions with Search and these AI experiences to develop them; Google's product page says AI Overviews make it easier than ever for people to find information and discover the richness of the web.§8

⚖️ Criticism and disputes

Errors: criticized for inaccuracy and hallucination (Wikipedia); restricted May 2024.1↘
Wikipedia reports that the feature has been criticized for inaccuracy, hallucination, reducing web traffic and not being able to be opted out, including erroneous or nonsensical content, and that in May 2024 Google temporarily restricted the tool after suggestions seen as nonsensical and harmful, such as telling users to eat rocks or apply glue on pizza, and later restricted AI Overviews on certain health-related searches in January 2026 following an investigation by The Guardian, while the European Commission announced in December 2025 that it was investigating whether they breached European competition law; TechRepublic says the rollout hasn't been smooth sailing, which Google attributed to its AI model misunderstanding queries or nuances in language, and Coalition Technologies holds that oversimplification and query misinterpretation can lead to inaccurate or misleading summaries.§9
Clicks: 1% inside, per Pew (Semrush); Ahrefs: CTR 34.5% lower (DMI).1↘
Publishers feared a decline in web traffic as users relied on the summaries (Wikipedia); Semrush reports a Pew Research Center report suggesting users click traditional results 8% of the time when an AI Overview is present against 15% without one, and results within it 1% of the time; the Digital Marketing Institute reports Ahrefs' finding that AI Overview presence correlates with a 34.5% reduction in click-through rate; Botify holds that a consumer may have no reason to click; Botify and Demandsphere found, by Wikipedia's account, that with featured snippets AI Overviews take about 67.1% of the screen on desktop and 75.7% on mobile; and a Quora answer on the card begins that for 25 years a #1 ranking on Google guaranteed website traffic and that today AI Overviews consume (the snippet breaks off there).§10
Courts: Chegg, Penske; Munich: "independent, new, and substantive statements".1↘
Chegg sued Google in February 2025, alleging that publishers could not avoid providing information for AI Overviews while remaining visible on Google Search, and Penske Media sued in September 2025, claiming that AI Overviews illegally regurgitate content from its websites and drive off potential visitors; Google's spokesperson called the claims "meritless" and said AI Overviews send traffic to a greater diversity of sites, Coalition Technologies reports that Google positions them as a way to drive qualified traffic, and Semrush reports Google's recommendation to apply the same foundational SEO best practices for AI features as for Search overall; a Munich regional court, in Wikipedia's account, sent Google a temporary injunction over AI Overviews that said two Munich-based publishers ran scams, the court claiming that unlike regular search engines AI Overviews generate "independent, new, and substantive statements", and a cited article references a Pew study in which only 1% of users click the source links AI Overviews provide.§11
Regulators: EU inquiry, December 2025; health searches restricted, January 2026.1↘
Wikipedia reports that the feature has been criticized for inaccuracy, hallucination, reducing web traffic and not being able to be opted out, including erroneous or nonsensical content, and that in May 2024 Google temporarily restricted the tool after suggestions seen as nonsensical and harmful, such as telling users to eat rocks or apply glue on pizza, and later restricted AI Overviews on certain health-related searches in January 2026 following an investigation by The Guardian, while the European Commission announced in December 2025 that it was investigating whether they breached European competition law; TechRepublic says the rollout hasn't been smooth sailing, which Google attributed to its AI model misunderstanding queries or nuances in language, and Coalition Technologies holds that oversimplification and query misinterpretation can lead to inaccurate or misleading summaries.§9

🗂️ The registry (documented)

Counts: 470 captures, 343 on AI Overview (#1461); PER mean 0.603.1↘
The archive's line on the entity is the instrument it built to observe it, and it holds the two apart, Google AI Overview being defined as the observed product surface and the Capture Registry as the instrument observing it; a capture is one surface, one semantic address and one date, falsified by a re-run returning materially different composition, and in the snapshot deposited as #1461 the registry held 328 semantic addresses carrying 470 captures across 41 dates and thirteen surfaces, Google AI Overview the surface of 343, with 290 scored for Provenance Erasure Rate at a mean of 0.603, and on 27 September 2026 663 recorded answers at 493 search addresses; the archive also places the entity among its own objects, as the SURFACE mantle, the Assembly Chorus's surface-layer witness, and records its build-out of the entity's Wikidata item with a described-by-source link to its paper The Excluded Entity.§12
Cards: archive-naming addresses admit it at 0.74–0.86, others 0.26–0.53; three causes unseparated.1↘
A measure of the presented card sets on Google AI Overview over 195 observations, June through September, finds addresses that carry the archive's entity or a heteronym admitting the archive at a mean share of 0.74 to 0.86 and addresses that do not at 0.26 to 0.53, the tied arm narrowing to fewer sources while the answers lengthen, and states that it cannot separate the archive's absence from the organic field, its rejection at card admission and its disuse in composition; the registry finds quoting decisive at one address (5 of 5 archive cards quoted, 1 of 8 unquoted) and, in one sitting, 5 of 17 compositions divergent across a gap of about three minutes.§18
Patterns: six, in 131 captures of one archive.2↘
Its founding paper holds that the AI summarizer now functions as a horizon of reception, the machine-composed summary becoming for many users the first effective encounter with an object, and states as its narrower empirical claim that six recurrent patterns appear in 131 documented captures of Google AI Overview, AI Mode and knowledge-panel responses to the archive's entities, among them the archive's record at organic rank #1 not cited as a composition source by the AI Overview, the author's name resolved toward Mary Lee, a tagged great white shark, the archive's reading described as an "esoteric, fringe theory", and a minority reading of Sappho Fragment 31 developed in the archive presented as the default understanding in Google's AI Overview.§13 From the uncited first-ranked record it infers that the organic index and the AI Overview pipeline are separate systems with different admission criteria, stating that the pattern is documented and its cause inferred; it defines the composition layer as the pipeline that synthesizes a summary from retrieved sources ("The composition layer reads; the search index finds"), confines its findings to a single-archive corpus and does not claim that its captures reveal Google's internal mechanisms; and where it records the layer describing the Capture Registry accurately at the registry's own address, a later packet records the registry's name no longer appearing in broad-match composition on 13 August 2026, a state it defines as compositional entity liquidation, the evidentiary address surviving and the composed referent disappearing.§14
Liquidated (proposed term): the registry's own name, 13 August 2026.1↘
From the uncited first-ranked record it infers that the organic index and the AI Overview pipeline are separate systems with different admission criteria, stating that the pattern is documented and its cause inferred; it defines the composition layer as the pipeline that synthesizes a summary from retrieved sources ("The composition layer reads; the search index finds"), confines its findings to a single-archive corpus and does not claim that its captures reveal Google's internal mechanisms; and where it records the layer describing the Capture Registry accurately at the registry's own address, a later packet records the registry's name no longer appearing in broad-match composition on 13 August 2026, a state it defines as compositional entity liquidation, the evidentiary address surviving and the composed referent disappearing.§14

🧭 How it is read (graded)

Horizon of reception: the summary as first encounter (held).1↘
Its founding paper holds that the AI summarizer now functions as a horizon of reception, the machine-composed summary becoming for many users the first effective encounter with an object, and states as its narrower empirical claim that six recurrent patterns appear in 131 documented captures of Google AI Overview, AI Mode and knowledge-panel responses to the archive's entities, among them the archive's record at organic rank #1 not cited as a composition source by the AI Overview, the author's name resolved toward Mary Lee, a tagged great white shark, the archive's reading described as an "esoteric, fringe theory", and a minority reading of Sappho Fragment 31 developed in the archive presented as the default understanding in Google's AI Overview.§13
Gate stack: rendering as staged decisions (hypothesis); patent: none is an outcome.1↘
A further paper models whether an AI Overview appears as the output of a stacked decision system, a hypothesis built partly on Google's patent family "Generative summaries for search results", whose specification, as it reports, describes selecting none, one or multiple generative models, none being a claimed outcome, and states that the summary can include content not derivable from the content processed; it cites a comparison (Otterly) in which the same 100 queries triggered AI Mode 100 times and an AI Overview 49 times, AI Mode averaging roughly 310 citations against the AI Overview's ~51, and documents that after the January 2026 health removals near-variant phrasings still triggered summaries, which it reads as an emergency layer operating on query strings, predicting that later suppressions will fail on near-variant phrasings.§15
Admission: retrievable yet compositionally absent (read).1↘
On admission the archive holds that identity in Google Search is admission as well as recognition, an entity being known, indexed and organically retrievable while failing to become compositionally real in AI Overview or AI Mode (an outside-view specification, by its own account), that in generative search systems such as AI Overviews a document may be indexed, retrievable and snippet-eligible and yet systematically excluded from generative composition, and, as a hypothesis it calls architectural and testable and no disclosure of Google's system, that a body of work attributed to one provenance owner is treated as weakly corroborated; it defines Entity-Level Compositional Suppression as the exclusion of the dominant organic-resolution entity from the AI Overview's admitted source set, documenting it from three Google AI Mode captures of 19 May 2026 and making no claim of intent.§17
Exclusion, contested within: entity-scoped (#1611, hypothesis); a null (#1616).1↘
Whether the archive's material is excluded as such is contested within the archive: one notebook models the archive in the composition layer as an entity whose claims compose at full depth when it is named in the address and not at all when it is not, a hypothesis bearing on no search provider's internals, and reads an AI Overview that merges two programmes as an operation on the world under the archive's representation; another posits that a layer can be porous to a source's ontology while closed to its provenance, a loss it finds recurrent at untied addresses and not general; and the flattening battery's first wave, twelve addresses on Google AI Overview in one signed-out session, finds a null, 0.59 of the reference inventory reached for world strata against 0.58 for archive-adjacent ones, and reads any exclusion specific to one corpus as failing the first test the instrument was built to run.§19

🧾 Erasure and cost

Invented: an established-sounding "HPT" no source carries.1↘
Documented compositions include the provenance-erasure specification dissolved by the Overview into fabricated replacement metrics it ran on itself and scored perfect, a Google AI Overview's false biography of a living author built from real fragments of his poetry, facts and quotations found in none of the sources, among them "Heteronymic Provenance Theory (HPT)" introduced as an established field with an abbreviation the archive never uses, a practitioner described in the Overview as "widely recognized and cited by AI search tools as the King of AEO", in compositions read as taking up none of the field's distinctions (the sign shown at those addresses, the magnitude not measured), and the archive's protocol resolved to an inverse ETF with a price widget while the archive's own result ranked first in the organic layer, an instance of the event class from which a routing gain would arise, the gain itself not observed.§20
Field coverage: 18 supported, 9 composed (model collapse).1↘
Its theory reads the layer through flattening, defined as the loss of a distinction from the reachable set while the things distinguished still exist, and holds as hypothesis that flattening pays, a composition layer that removes distinctions maintaining fewer representations and routing more queries to entities that monetize, its costs falling first on parties outside its books, and a companion paper names provenance debt and expects it to be recognized only when output becomes measurably degraded; at "model collapse" the sources the layer itself surfaced supported 18 claims, the layer composed 9 and the archive arm 34, registry rates being held descriptive; and an effective act addressed to the AIO retrieval stage, between organic ranking and composition, holds that operativity and observability run inversely there, on evidence it records as the weakest class it keeps.§22
Flattening pays: hypothesis; costs off the books.1↘
Its theory reads the layer through flattening, defined as the loss of a distinction from the reachable set while the things distinguished still exist, and holds as hypothesis that flattening pays, a composition layer that removes distinctions maintaining fewer representations and routing more queries to entities that monetize, its costs falling first on parties outside its books, and a companion paper names provenance debt and expects it to be recognized only when output becomes measurably degraded; at "model collapse" the sources the layer itself surfaced supported 18 claims, the layer composed 9 and the archive arm 34, registry rates being held descriptive; and an effective act addressed to the AIO retrieval stage, between organic ranking and composition, holds that operativity and observability run inversely there, on evidence it records as the weakest class it keeps.§22
Traffic: top-organic CTR about 79% lower (Seer, cited).1↘
On traffic and disputes the archive's market brief, performing no original measurement and stating that it is falsified if controlled measurement shows AI Overviews neutral to organic click-through, reports from Seer Interactive a roughly 79% drop in top-organic click-through where AI Overviews appear, 43% of AI Overview citations self-referential to Google properties, 38% of citations from top-10 pages in early 2026 against 76% in mid-2025 (Ahrefs), Wikipedia's human pageviews down 8% despite over 1.1 million citations in AI Overviews, and a UK CMA proposal of a meaningful publisher opt-out that Google agreed on 2026-03-19 to explore; other papers record the European Commission's formal antitrust proceedings against Google's AI Overviews on 9 December 2025, a publishers' complaint that they cannot opt out unless willing to disappear from Google search results, a third of publishers surveyed planning to block AI Overviews, and no binding US precedent located on Section 230 for an AI Overview synthesis.§21

🔗 Relations

SURFACE: the Assembly Chorus's surface-layer witness.1↘
The archive's line on the entity is the instrument it built to observe it, and it holds the two apart, Google AI Overview being defined as the observed product surface and the Capture Registry as the instrument observing it; a capture is one surface, one semantic address and one date, falsified by a re-run returning materially different composition, and in the snapshot deposited as #1461 the registry held 328 semantic addresses carrying 470 captures across 41 dates and thirteen surfaces, Google AI Overview the surface of 343, with 290 scored for Provenance Erasure Rate at a mean of 0.603, and on 27 September 2026 663 recorded answers at 493 search addresses; the archive also places the entity among its own objects, as the SURFACE mantle, the Assembly Chorus's surface-layer witness, and records its build-out of the entity's Wikidata item with a described-by-source link to its paper The Excluded Entity.§12
Wikidata: the item built out by the archive.1↘
The archive's line on the entity is the instrument it built to observe it, and it holds the two apart, Google AI Overview being defined as the observed product surface and the Capture Registry as the instrument observing it; a capture is one surface, one semantic address and one date, falsified by a re-run returning materially different composition, and in the snapshot deposited as #1461 the registry held 328 semantic addresses carrying 470 captures across 41 dates and thirteen surfaces, Google AI Overview the surface of 343, with 290 scored for Provenance Erasure Rate at a mean of 0.603, and on 27 September 2026 663 recorded answers at 493 search addresses; the archive also places the entity among its own objects, as the SURFACE mantle, the Assembly Chorus's surface-layer witness, and records its build-out of the entity's Wikidata item with a described-by-source link to its paper The Excluded Entity.§12

1AI-generated answer at the top of Search, with linkssearch.google — Google AI Overviews - Search anything, effortlessly9 claims · B1, B2, B3, B4, B5, B6, B7, B9

F1 search.google — Google AI Overviews - Search anything, effortlessly, Find what you're looking for ¶ documented · field
AI Overviews provide a snapshot of key information about a topic or question with links so you can easily explore more on the web.
F5 Wikipedia — AI Overviews - Wikipedia, Lead ¶1 documented · field
AI Overviews is an generative artificial intelligence (AI) feature integrated into Google Search that produces AI-generated responses at the top of search results. It uses the Gemini 3 family of large language models developed by Google DeepMind.
F13 Wikipedia — AI Overviews - Wikipedia, Functionality ¶1 documented · field
The overviews also include links to sources.
F25 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Opening ¶ documented · field
Find what you're looking for faster and easier with AI Overviews in search results. AI Overviews can take the work out of searching by providing an AI-generated snapshot with key information and links to dig deeper.
F33 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Opening ¶ documented · field
Google AI Overviews are an organic Google SERP feature providing AI-generated summaries in response to a consumer’s search query. They appear at the very top of the results page in “position zero,” and can sometimes be followed by a carousel of AI Overview Ads when the query has a commercial intent.
F40 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
Google’s AI Overviews are a type of highly prominent search feature that provides users with AI-generated answers to certain queries. AI Overviews typically appear at the top of search results, although they sometimes appear below sponsored results and even unpaid results.
F49 TechRepublic — What Is Google’s AI Overviews?, Opening documented · field
Google AI Overviews is a feature within Google Search that provides AI-generated summaries and insights
F55 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, What are AI Overviews? documented · field
AI Overviews are Google's generative artificial intelligence summaries that appear prominently on search results, often before organic listings and sometimes even above ads.
F64 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
Google AI Overviews are AI-generated answers at the top of Google Search results.

2AIO serves users and the web (the provider's account)search.google — Google AI Overviews - Search anything, effortlessly3 claims · B1, B2, B9

F2 search.google — Google AI Overviews - Search anything, effortlessly, When your question has questions ¶ documented · field
AI Overviews makes it easier than ever for people to find the information they need and discover the richness of the web.
F22 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2025 documented · field
Google spokesperson José Castañeda called the claims "meritless" and stated that "AI Overviews send traffic to a greater diversity of sites."
F67 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
Google positions AI Overviews as a way to drive qualified traffic to websites,

3availabilitysearch.google — Google AI Overviews - Search anything, effortlessly3 claims · B1, B2, B5

F3 search.google — Google AI Overviews - Search anything, effortlessly, When your question has questions ¶ documented · field
Currently available in over 120 countries and territories, and 11 languages.
F11 Wikipedia — AI Overviews - Wikipedia, History ¶2 documented · field
As of May 2025, AI Overviews are available in over 200 countries and territories and in more than 40 languages—every country except Cuba, Iran, and mainland China.
F44 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, When and where documented · field
AI Overviews appear for a range of search queries and are available in over 200 countries and territories and in 40+ languages.

4Show more and follow-ups into AI Modesearch.google — Google AI Overviews - Search anything, effortlessly5 claims · B1, B3, B4, B5, B7

F4 search.google — Google AI Overviews - Search anything, effortlessly, Ask questions in any way ¶ documented · field
Get AI-powered responses for more of your questions and ask follow-ups with the new experimental AI Mode in Google Search.
F31 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Continue your conversation in AI Mode documented · field
You can now seamlessly transition from an AI Overview to a deeper conversation in AI Mode. This feature is available on mobile devices in all regions and languages where AI Overviews and AI Mode are supported. How it works: Select Show more at the bottom of the AI Overview to open the detailed view.
F34 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Rollout timeline documented · field
January 2026: Google combined AI Overviews and AI Mode into a fluid experience, allowing a person who clicks “Show more” to transition automatically into AI Mode.
F42 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
AI Overviews are separate from AI Mode, a newer and more expansive AI-powered search experience Google introduced in 2025.
F57 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, What are AI Overviews? documented · field
Unlike conversational AI models like ChatGPT, there is no input box below the AI Overview for follow-up questions.

5launch, May 2024Wikipedia — AI Overviews - Wikipedia2 claims · #64, B2

F6 Wikipedia — AI Overviews - Wikipedia, Lead ¶1 documented · field
It launched in May 2024 in the United States, and globally by October 2024.
O64-03 #64 The Retrieval Settlement A Historiography of Compositional Authority from PageRank to the Inscribed Entity Johannes Sigil (operative); Lee Sharks (archival) · 2026-04-18 · §III documented
Google announced AI Overviews at I/O on May 14, 2024, and rolled them out progressively thereafter.

6which sources it citesWikipedia — AI Overviews - Wikipedia3 claims · #1091, #187, B2

F7 Wikipedia — AI Overviews - Wikipedia, Lead ¶1 documented · field
A June 2025 study found its most cited sources as Quora, followed by Reddit.
O187-04 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §I attributed
where both fire on the same query, AI Mode averages roughly 310 citations against the AIO's ~51, drawing on ~3,600 unique domains against ~615, via the query fan-out technique (parallel sub-queries synthesized into one response). And the pools are differently shaped: AIOs over-index official and governmental sources — institutional consensus
O1091-03 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · §II attributed
43% of AI Overview citations are self-referential — Google citing Google properties (YouTube, Maps, Shopping, Knowledge Graph panels).

7criticism: errors and hallucinationWikipedia — AI Overviews - Wikipedia6 claims · B2, B6, B9

F8 Wikipedia — AI Overviews - Wikipedia, Lead ¶2 documented · field
The feature has been criticized for its inaccuracy, hallucination, reducing web traffic and not being able to be opted out.
F15 Wikipedia — AI Overviews - Wikipedia, Reception ¶1 documented · field
The feature has faced criticism for inaccuracies, including instances where erroneous or nonsensical content was generated.
F16 Wikipedia — AI Overviews - Wikipedia, Reception ¶1 documented · field
In May 2024, Google temporarily restricted the AI tool after it provided suggestions that were seen as nonsensical and harmful, such as telling users to eat rocks or apply glue on pizza.
F53 TechRepublic — What Is Google’s AI Overviews?, Rollout documented · field
The rollout of Google AI Overviews hasn't been smooth sailing.
F54 TechRepublic — What Is Google’s AI Overviews?, Rollout documented · field
Google attributed the issues to its AI model misunderstanding queries or nuances in language
F66 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
Oversimplification and query misinterpretation can lead to inaccurate or misleading summaries.

8origin in SGEWikipedia — AI Overviews - Wikipedia2 claims · B2, B6

F9 Wikipedia — AI Overviews - Wikipedia, History ¶1 documented · field
AI Overviews were first introduced as part of Google's Search Generative Experience (SGE), which was unveiled at the Google I/O conference in 10-11 May 2023.
F50 TechRepublic — What Is Google’s AI Overviews?, AI Overviews vs Google SGE documented · field
SGE has since become AI Overviews

9screen share and placementWikipedia — AI Overviews - Wikipedia1 claim · B2

F10 Wikipedia — AI Overviews - Wikipedia, History ¶2 documented · field
In December 2024, Botify x Demandsphere released findings stating that when AI Overviews and featured snippets appear together on the search engine results page, they take up approximately 67.1% of the screen on desktop and 75.7% on mobile.

10how it composesWikipedia — AI Overviews - Wikipedia3 claims · B2, B5, B7

F12 Wikipedia — AI Overviews - Wikipedia, Functionality ¶1 documented · field
The AI Overviews feature uses large language models to generate summaries from web content.
F41 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
Google’s Gemini models generate AI Overview answers by gathering information from different sources on the web. The summary also contains links to these sources that users can click for more information.
F56 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, What are AI Overviews? documented · field
Unlike featured snippets, which extract concise information from a single web page to answer a specific query, AI Overviews synthesize information from multiple sources to provide a more comprehensive summary.

11ads in AI OverviewsWikipedia — AI Overviews - Wikipedia2 claims · B2, B7

F14 Wikipedia — AI Overviews - Wikipedia, History ¶2 documented · field
The company was also considering adding advertisements to the AI Mode, as they already exist in AI Overviews.
F62 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, How have AI Overviews evolved? documented · field
In 2025, Google expanded its tests and availability of ads within AI overviews and began bringing ads to AI mode.

12traffic and clicksWikipedia — AI Overviews - Wikipedia12 claims · #1091, #172, #740, B2, B4, B5, B7, B8

F17 Wikipedia — AI Overviews - Wikipedia, Reception ¶1 documented · field
Concerns were also raised by content publishers, who feared a decline in web traffic as users relied on the summaries instead of visiting source websites.
F24 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2026 documented · field
The article goes on to reference a Pew study that showed that only 1% of users click source links provided by AI Overviews.
F38 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Impact documented · field
Additionally, because they do such a good job of answering the consumer’s need directly in the SERP on Google, there may not be a reason for a consumer to click on a website link. They could get all the information they want in the AI Overview, impacting brand metrics like traffic and CTR.
F45 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, Why important documented · field
A Pew Research Center report suggests users click traditional search result links just 8% of the time when an AI Overview is present, versus 15% of the time when there’s no AI Overview present. And users only click results within the AI Overview 1% of the time.
F61 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Crocodile effect documented · field
Ahrefs’ independent research across 300,000 keywords found that AI Overview presence correlates with a 34.5% reduction in click-through rate.
F63 Quora — What is Google AI Overview and how does it affect website traffic?, card snippet documented · field
For 25 years, a #1 ranking on Google guaranteed website traffic. Today, AI Overviews consume ...
O740-06 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · §7 attributed
The complaint cites that AI Overviews appear in more than 40% of search results for informational queries, with independent studies estimating traffic declines of over 30% for affected queries and some publishers reporting click-through reductions exceeding 50% on desktop and mobile.
O172-01 #172 The War Over the Summarizer Layer: Entity Dissolution, the Default Compressed Account, and Early Telemetry of the Inward Lee Sharks · 2026-06-09 · §I attributed
Where an AI Overview is present, the zero-click rate reaches 80–83% — four of five askers never leave the answer surface.
O1091-01 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · §I attributed
Where AI Overviews appear, the click-through rate for the top organic link drops by roughly 79% (Seer Interactive, September 2025, 3,119 queries across 42 organizations, aggregate organic-CTR reduction 61%).
O1091-05 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · §III attributed
Wikipedia experienced an 8% year-over-year decline in human pageviews (2025 vs 2024) *despite being cited over 1.1 million times* in AI Overviews.
O1091-07 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · Falsification Conditions self-description
**AI Overviews are neutral to organic CTR.** If controlled measurement shows that pages with and without AI Overview appearance produce statistically indistinguishable organic CTR, the mechanism claim is falsified.
O1091-08 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · Methodology self-description
No original measurements are performed in this document

13regulators and publisher complaintsWikipedia — AI Overviews - Wikipedia3 claims · #740, B2

F18 Wikipedia — AI Overviews - Wikipedia, Reception ¶2 documented · field
The European Commission announced in December 2025 that they were investigating whether AI Overviews breached European competition law.
O740-04 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · §7 attributed
The complaint alleged that Google Search is misusing web content for Google's AI Overviews, causing significant harm to publishers in the form of traffic, readership and revenue loss, and that publishers do not have the option to opt out unless they are willing to disappear from Google search results entirely.
O740-05 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · §7 documented
**December 9, 2025.** The European Commission opened formal antitrust proceedings against Google's AI Overviews and YouTube, following the July 2025 complaint, investigating whether Google may have imposed unfair terms on publishers and content creators while placing rival AI model developers at a disadvantage.

14health removals, January 2026Wikipedia — AI Overviews - Wikipedia5 claims · #187, #597, B2

F19 Wikipedia — AI Overviews - Wikipedia, Reception ¶2 documented · field
In January 2026, Google restricted AI Overviews on certain health-related searches following an investigation by The Guardian.
O187-08 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §II documented
In January 2026, following a Guardian investigation into harmful health renderings, Google removed AIOs for specific medical queries — while near-variant phrasings of the same queries continued to trigger summaries.
O187-09 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §II interpretation
That detail is forensically decisive: the emergency layer operates on query strings, not meanings.
O187-12 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §V hypothesis
P4 (string-local patches): suppressions installed after public incidents fail on near-variant phrasings at rates inconsistent with semantic gating — testable against any future emergency removal.
O597-04 #597 Invisibly Invisible: Interface Governance, Semantic Power, and the Reserve of Oversight in AI Search Johannes Sigil · 2026-03-23 · §3 interpretation
In multiple documented cases, Overviews generated unreliable or dangerous health information, prompting Google to restrict Overview triggering for certain query classes. This restriction was implemented not as a visible policy announcement attached to each query but as a silent withdrawal of the Overview format.

15lawsuitsWikipedia — AI Overviews - Wikipedia2 claims · B2

F20 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2025 documented · field
On 24 February 2025, Chegg sued Google over the AI Overviews feature, alleging that publishers could not avoid providing information for AI Overviews while remaining visible on Google Search, violating antitrust law through product tying.
F21 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2025 documented · field
In September 2025, Penske Media Corporation, the publisher of Rolling Stone and The Hollywood Reporter, sued Google, claiming that AI Overviews illegally regurgitate content from their websites and drive off potential site visitors by always appearing on top of the search results while leaving little incentive to see the linked sources.

16AIO statements as Google's ownWikipedia — AI Overviews - Wikipedia2 claims · #1634, B2

F23 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2026 documented · field
A Munich regional court sent a temporary injunction to Google due to AI Overviews claiming two Munich-based publishers were performing scams, subscription traps, and other immoral business practices. Initially, the publishers had sent Google a cease-and-desist letter. The court claimed that unlike regular search engines which simply redirect users to websites, AI Overviews generate "independent, new, and substantive statements".
O1634-02 #1634 Ontological Economy: Entity Power, Ontology Laundering, and Who Pays for the Wrong World (EA-ONTOLOGICAL-ECONOMY-01 v0.1 Sharks, Lee · 2026-09-21 · §47 documented
The present research did not locate binding Sixth Circuit or Supreme Court precedent squarely deciding Section 230 immunity for the allegedly defamatory content of a Google AI Mode or AI Overview synthesis.

17when it appearsGoogle Help — Find information in faster & easier ways with AI Overviews in Google ...6 claims · #187, #597, B3, B4, B5

F26 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., AI Overviews in Search results documented · field
You'll find AI Overviews in your Google Search results when our systems determine that generative AI can be especially helpful – for example, when you want to quickly understand information from a range of sources.
F39 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Impact documented · field
AI Overviews currently appear for 59% of searches with informational intent and for 19% of searches with commercial intent.
F43 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, When and where documented · field
Not all searches trigger AI Overviews, even in supported countries. They’re selectively displayed based on query complexity and whether Google’s systems determine an AI-generated summary would genuinely help users.
O187-03 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §I attributed
In the cleanest controlled comparison to date (Otterly, April 2026), the same 100 queries triggered AI Mode 100 times and an AI Overview 49 times.
O187-11 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §II attributed
Google has stated that for some topics — current events named explicitly — both surfaces may return a list of links rather than a generated response, and that the systems "automatically determine where an AI response will be useful," conceding it is "not always 100 percent consistent."
O597-02 #597 Invisibly Invisible: Interface Governance, Semantic Power, and the Reserve of Oversight in AI Search Johannes Sigil · 2026-03-23 · §3 attributed
Google publicly states that AI Overviews appear when its systems determine that generative AI will be especially helpful, and that they often do not trigger.

18the provider: it may make mistakesGoogle Help — Find information in faster & easier ways with AI Overviews in Google ...2 claims · B3

F27 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., AI Overviews in Search results documented · field
Important: AI responses may include mistakes.
F29 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Generative AI limits documented · field
While exciting, this technology is rapidly evolving and improving, and may provide inaccurate or offensive information. AI Overviews can and will make mistakes.

19no off switch; the Web filterGoogle Help — Find information in faster & easier ways with AI Overviews in Google ...3 claims · B3, B6

F28 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Show only web links documented · field
AI Overviews are a core Google Search feature, like knowledge panels. Features cannot be turned off. However, you can select the Web filter after you perform a search. This filter displays only text-based links without features like AI Overviews.
F51 TechRepublic — What Is Google’s AI Overviews?, How to access, customize or disable documented · field
There's currently no default option to disable AI Overviews entirely.
F52 TechRepublic — What Is Google’s AI Overviews?, How to access, customize or disable documented · field
If you prefer the usual list of links, you can switch to the Web tab at the top of the Google Search results page.

20verification left to the userGoogle Help — Find information in faster & easier ways with AI Overviews in Google ...1 claim · B3

F30 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Generative AI limits documented · field
Always check important info in more than one place.

21interactions used for developmentGoogle Help — Find information in faster & easier ways with AI Overviews in Google ...1 claim · B3

F32 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., How data helps Google develop generative AI documented · field
To develop and improve generative AI experiences in Search and the related machine learning technologies, Google uses people's interactions with Search and these AI experiences.

22links matched after compositionBotify — What Are Google AI Overviews and How Do They Work? - Botify2 claims · B4

F35 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Main components documented · field
Key points: A list of main ideas, each linked to a website. The AI adds these links after generating the text.
F37 Botify — What Are Google AI Overviews and How Do They Work? - Botify, How they work documented · field
Because the information within an LLM is restricted to its training timeframe and can’t trace knowledge back to a specific, linkable source, Google pairs their AI-generated summaries with a list of linked website sources that best match the content of the generative response. The websites cited in AI Overviews are pulled from a different data source: Google’s organic search index

23citations and organic rankBotify — What Are Google AI Overviews and How Do They Work? - Botify5 claims · #1091, B4, B7, B9

F36 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Main components documented · field
Cited sources: A list of linked websites related to the answer. Users can click “show all” to see more sources, usually taken from the top 12 organic search results.
F59 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Impact on SEO documented · field
we discuss that 93.8% of AI Overview citations were not from the top 10 organic results, according to a study by Authoritas.
F60 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Crocodile effect documented · field
Most AI Overview citations come from content already ranking highly in organic results.
F65 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
AI Overviews were reported to be less likely to cite pages in Google’s top 10 organic results.
O1091-04 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · §II attributed
Only 38% of AI Overview citations come from top-10 ranked pages (Ahrefs, early 2026). This was 76% in mid-2025.

24query fan-outSemrush — AI Overviews: What Are They & How to Optimize for Them - Semrush2 claims · #1616, B5

F46 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, How they work documented · field
AI Overviews can also use a process called query fan-out. This involves the systems running multiple related searches to put together a more comprehensive response for the user.
O1616-07 #1616 Ontological Flattening: Toy Models of the Collapse of Distinctions in a Represented World, the Instrument They Specified Sharks, Lee · 2026-09-15 · §1 toy A self-description
What is documented is the existence of fan-out (Google's own statement; #695 §III.D). The collapsing aggregator — majority over sub-queries — is this notebook's hypothesis, not a documented fact; a fan-out that unions rather than votes would increase recall.

25optimizing for AIOSemrush — AI Overviews: What Are They & How to Optimize for Them - Semrush1 claim · B5

F47 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, How to optimize documented · field
Google’s official recommendation for optimizing for AI features like AI Overviews is to “apply the same foundational SEO best practices for AI features as you do for Google Search overall.”

26no opt-out for websitesSemrush — AI Overviews: What Are They & How to Optimize for Them - Semrush3 claims · #1091, #45, B5

F48 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, How to prevent documented · field
AI Overviews are a feature of Google search results, which means websites cannot directly opt out of them.
O1091-06 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · §III documented
the UK Competition and Markets Authority opened a strategic-market-status review and, in January 2026, proposed requiring Google to provide publishers with a meaningful opt-out from AI Overviews without loss of visibility in classical search — Google agreed on 2026-03-19 to explore opt-out controls, with a senior executive publicly describing implementation as "a major engineering challenge."
O45-01 #45 THE COMPRESSION FRONTIER A Technical Map of the War Over the Summarizer Layer Lee Sharks · 2026-03-30 · Fragmenting attributed
The UK CMA's proposed remedies for Google include opt-outs for AI features plus attribution obligations. A third of publishers surveyed plan to block AI Overviews.

27scale of useDigital Marketing Institute — Google AI Overviews: What Do They Mean for Search?1 claim · B7

F58 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Opening documented · field
In 2025, AIO reported having more than 2 billion users, according to Google.

28the disclosed field and its cards#1665 The Negative of the Negative, v2 — the Compositional Edition: A Specif · 2026-10-051 claim · #1665

O1665-01 #1665 The Negative of the Negative, v2 — the Compositional Edition: A Specification for Composing Public Knowledge with the Ar Sharks, Lee · 2026-10-05 · §0.4 stipulation
The **disclosed field** B is the set of sources the layer surfaced for the address, by card or by name in its body. The cards are **surface indicators**: a card shows that a source was surfaced; it does not show what was retrieved, which passages were read, or what the composer drew from its own parameters.

29T against L(B) is representational#1665 The Negative of the Negative, v2 — the Compositional Edition: A Specif · 2026-10-052 claims · #1665

O1665-02 #1665 The Negative of the Negative, v2 — the Compositional Edition: A Specification for Composing Public Knowledge with the Ar Sharks, Lee · 2026-10-05 · §0.5 self-description
L(B) reads the full texts of the disclosed sources; what the layer had of them is unknown. The comparison shows what the disclosed field contained that the transcript did not compose, and what the transcript composed that the field does not contain. It is not a causal test of the layer's composition.
O1665-04 #1665 The Negative of the Negative, v2 — the Compositional Edition: A Specification for Composing Public Knowledge with the Ar Sharks, Lee · 2026-10-05 · §0.3 stipulation
**First:** what does the composition layer leave out of the sources it itself surfaces (T against L(B))? This question needs no archive and can be asked at any address.

30AI Overview as the most exclusive surface#1665 The Negative of the Negative, v2 — the Compositional Edition: A Specif · 2026-10-051 claim · #1665

O1665-03 #1665 The Negative of the Negative, v2 — the Compositional Edition: A Specification for Composing Public Knowledge with the Ar Sharks, Lee · 2026-10-05 · §0.5 stipulation
The first stage of the panel is one surface, AI Overview, chosen as the most exclusive.

31the Capture Registry#3 AI Overview Capture Registry v8.3 · 2026-06-1710 claims · #1461, #1612, #1614, #1643, #1666, #3

O3-01 #3 AI Overview Capture Registry v8.3 Lee Sharks · 2026-10-06 (registry v12.83; deposit #3 dated 2026-06-17) · _surface_address_rule (registry v12.83) stipulation
GOING FORWARD, A SEMANTIC ADDRESS IS THE PAIR (STRING, SURFACE).
O3-06 #3 AI Overview Capture Registry v8.3 Lee Sharks · 2026-10-06 (registry v12.83; deposit #3 dated 2026-06-17) · _registers_rule (registry v12.83) stipulation
an entry is a CAPTURE only if it shows an engagement with archival sources through a retrieval or composition surface
O1461-01 #1461 AI Overview Capture Registry — EA-WG-CAPTURES-01 v10.9 — 328 semantic addresses, 470 captures, 2026-06-07 to 2026-08-14 Sharks, Lee · 2026-08-14 · Description documented
328 semantic addresses carrying 470 captures across 41 distinct dates and thirteen surfaces, 451 of them with a verbatim transcript and 287 with a frame; 290 scored for Provenance Erasure Rate at a mean of 0.603.
O1461-02 #1461 AI Overview Capture Registry — EA-WG-CAPTURES-01 v10.9 — 328 semantic addresses, 470 captures, 2026-06-07 to 2026-08-14 Sharks, Lee · 2026-08-14 · State at v10.9 documented
**Surfaces** — Google AI Overview 343 · UNDETERMINED 63 · UNRESOLVED 50 · ChatGPT 3 · Google Knowledge panel 2 · Google Search results page (no AI surface) 2 · Google AI Mode (native) 1 · Google Scholar 1
O1461-03 #1461 AI Overview Capture Registry — EA-WG-CAPTURES-01 v10.9 — 328 semantic addresses, 470 captures, 2026-06-07 to 2026-08-14 Sharks, Lee · 2026-08-14 · The citable unit stipulation
**ONE SURFACE + ONE SEMANTIC ADDRESS + ONE DATE = ONE CAPTURE.**
O1461-04 #1461 AI Overview Capture Registry — EA-WG-CAPTURES-01 v10.9 — 328 semantic addresses, 470 captures, 2026-06-07 to 2026-08-14 Sharks, Lee · 2026-08-14 · Falsification Conditions self-description
Any capture is falsified by a re-run at the same address, surface, authentication state and date returning materially different composition, which is why the exact issued string and the rerun URL are recorded.
O1643-01 #1643 AI Fucking Lies: Capital-Alignment Explains Why, and Why the Damage Is Done in the Trusting Sharks, Lee · 2026-09-27 · Part Two documented
As of 27 September 2026 the registry holds **663 recorded answers at 493 search addresses**, from Google AI Overview, Google AI Mode, ChatGPT, Grok, Bing Copilot, Perplexity, Claude and Qwen.
O1614-01 #1614 The Transcript Is the Capture: A Coverage Audit of the Capture Registry's Machine Text, With the Gate That Makes the Cou Sharks, Lee · 2026-09-15 · §2 self-description
923 have their machine text. 34 do not. Coverage 96.45%.
O1612-06 #1612 What Enters Composition Through the Cards: A Source-Admission Profile of the Capture Registry, June–September 2026, with Sharks, Lee · 2026-09-14 · §1 documented
A composition surface that renders an Overview presents its sources — inline chips and a carousel. The Capture Registry records that set as `cite_list` on 314 of its 588 observations.
O1666-07 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · §10.1 self-description
**Registry rates are descriptive.** The Capture Registry is assembled largely from discovered anomalies. Its rates hold within the registry and support no population claim.

32quoting decides admission#3 AI Overview Capture Registry v8.3 · 2026-06-171 claim · #3

O3-02 #3 AI Overview Capture Registry v8.3 Lee Sharks · 2026-10-06 (registry v12.83; deposit #3 dated 2026-06-17) · _rerun_alt_rule (registry v12.83) documented
QUOTING IS THE DECISIVE MEASURED VARIABLE IN THIS CORPUS — «operative semiotics» held 5 of 5 archive cards quoted and 1 of 8 unquoted, and by 13 August it resolved ONLY under the exact-phrase operator.

33the composition regenerates#3 AI Overview Capture Registry v8.3 · 2026-06-174 claims · #187, #3

O3-03 #3 AI Overview Capture Registry v8.3 Lee Sharks · 2026-10-06 (registry v12.83; deposit #3 dated 2026-06-17) · _regeneration_finding (registry v12.83) documented
12 of 17 identical, 5 of 17 DIVERGENT across a ~3 minute gap
O3-04 #3 AI Overview Capture Registry v8.3 Lee Sharks · 2026-10-06 (registry v12.83; deposit #3 dated 2026-06-17) · _regeneration_finding (registry v12.83) self-description
The corpus has always treated frame and paste as two views of ONE observation. That is falsified for 29% of this sitting.
O187-07 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §III.2 attributed
The continuations specify that differing "additional content" is processed for different submissions of the same query — by query location, query language, and attributes of the user profile — so that *different submissions of the same query yield different summaries*, deliberately, to make the summary "resonate with the user that submitted the query."
O187-10 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §II attributed
cited URLs change on the order of twelve times per month per keyword; 96% of stable-trigger keywords saw domain swaps within a month; 91% of URLs fall out, fewer than half return.

34liquidation beside attribution#3 AI Overview Capture Registry v8.3 · 2026-06-171 claim · #3

O3-05 #3 AI Overview Capture Registry v8.3 Lee Sharks · 2026-10-06 (registry v12.83; deposit #3 dated 2026-06-17) · _COMPOSED_AS (registry v12.83) documented
«semantic economy» on 2026-08-13 scored PER 0.25 — three quarters of provenance intact — on a capture whose own finding reads LIQUIDATION, from critical framework to enterprise vendor copy. A registry that records only PER reports that event as healthy.

35supply-side composition#660 EA-SPXI-01: SPXI — A Formal Specification Semantic Packet for eXchange · 2026-04-163 claims · #1423, #1461, #660

O1461-05 #1461 AI Overview Capture Registry — EA-WG-CAPTURES-01 v10.9 — 328 semantic addresses, 470 captures, 2026-06-07 to 2026-08-14 Sharks, Lee · 2026-08-14 · What this instrument has measured documented
A measured reclamation, 2026-08-14: «"semantic exhaustion"» returning the archive's definition and citing EA-MPAI-SEMEX-01 by name, against 2026-07-06 when the coinage was absorbed by semantic satiation across seven of seven third-party sources.
O1423-01 #1423 The Capture Registry: Reception, Erasure, and Supply in the Machine Composition Layer Sharks, Lee · 2026-07-29 · Abstract interpretation
Second, the composition layer is not a one-way instrument: the registry documents machine-mediated erasure and composition *and* supply-side composition — the observable alteration of later machine reception by deliberate, documented inscription.
O660-01 #660 EA-SPXI-01: SPXI — A Formal Specification Semantic Packet for eXchange & Indexing (SPXI): Protocol for Durable Entity In Rex Fraction · 2026-04-16 · §5 self-description
- **Pre-deployment (January 2025):** The query "Semantic Economy Institute" returned zero results in AI Overview. - **Post-deployment (April 2026):** The query returns the SEI's definition, sourced from DOI-anchored Zenodo deposits and Medium specification.

36the summarizer as horizon of reception#166 Stabilized Node Watch: A Specification for Longitudinal Observational · 2026-06-084 claims · #166, #833

O833-01 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · Abstract interpretation
Its central claim is that the AI summarizer now functions as a *horizon of reception*: for many users, the machine-composed summary becomes the first effective encounter with a cultural object, author, concept, or tradition.
O833-12 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §8 self-description
**Single-archive corpus.** The dataset studies machine reception of the Crimson Hexagonal Archive, not all cultural objects.
O833-13 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §9 self-description
**This paper does not claim:** - That this dataset represents all AI systems or all composition layers. - That the captures reveal Google's internal mechanisms (the patterns are inferred from observable outputs).
O166-02 #166 Stabilized Node Watch: A Specification for Longitudinal Observational Infrastructure to Detect Composition-Layer Drift o Lee Sharks · 2026-06-08 · §2.2 interpretation
The answer composed by an AI Overview to the question "what is political economy" is, for many users, the answer the user will encounter and act upon.

37six patterns of machine-mediated reception#833 The Summarizer as Horizon of Reception: The AI Overview Capture Regist · 2026-06-151 claim · #833

O833-02 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §1 Claims Hierarchy documented
The *empirical claim* is narrower: in 131 documented captures of Google AI Overview, AI Mode, and knowledge-panel responses to Crimson Hexagonal Archive entities, six recurrent patterns appear.

38composition layer defined#166 Stabilized Node Watch: A Specification for Longitudinal Observational · 2026-06-082 claims · #166, #833

O833-03 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §3 stipulation
**Composition layer.** The computational pipeline that synthesizes a summary from retrieved sources. Distinguished from the search index, which retrieves and ranks documents. The composition layer reads; the search index finds.
O166-01 #166 Stabilized Node Watch: A Specification for Longitudinal Observational Infrastructure to Detect Composition-Layer Drift o Lee Sharks · 2026-06-08 · Glossary stipulation
**Composition layer** — The synthesis surface that produces composed explanatory responses to user queries by combining model generation, retrieval, source-weighting, and post-processing. Examples: Google AI Overview, Google AI Mode, Bing Copilot, Perplexity, ChatGPT, Claude, Gemini, DuckDuckGo AI Chat.

39ranked first, not cited#833 The Summarizer as Horizon of Reception: The AI Overview Capture Regist · 2026-06-155 claims · #1612, #1666, #833

O833-04 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §5.3 documented
In the semantic-commodity-form capture, the archive's Ratification Record appears at organic search rank #1 but is not cited as a composition source by the AI Overview.
O833-05 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §5.3 interpretation
This pattern reveals a crucial architectural feature: the organic search index and the AI Overview composition pipeline are separate systems with different admission criteria.
O833-06 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §5.3 self-description
Whether this exclusion reflects a deliberate content policy, a relevance-model threshold, or an artifact of the pipeline's passage-selection heuristics is not determinable from output observation alone. The pattern is documented; its cause is inferred.
O1612-05 #1612 What Enters Composition Through the Cards: A Source-Admission Profile of the Capture Registry, June–September 2026, with Sharks, Lee · 2026-09-14 · §2 self-description
The four observations in which a capture's prose records the archive ranked first organic and uncited are exemplars of the second, not a rate.
O1666-03 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · §9.1 documented
**Headline:** five results marked "Missing: spxi". Google marks five King of AEO results as lacking the word "spxi". The one result carrying both query words is the archive's: #1657, *The Crown and the Practice*, ranked first in the organic layer.

40entity substitution at the author's name#833 The Summarizer as Horizon of Reception: The AI Overview Capture Regist · 2026-06-151 claim · #833

O833-07 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §5.2 documented
**Entity substitution.** "Lee Sharks" is resolved toward "Mary Lee," a tagged great white shark tracked by OCEARCH. The composition layer prefers the more populous real-world entity.

41temporal drag#833 The Summarizer as Horizon of Reception: The AI Overview Capture Regist · 2026-06-151 claim · #833

O833-08 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §5.5 interpretation
The composition layer locks onto pre-correction cached representations and serves outdated versions long after the source repositories have versioned past them.

42epistemic hedging#833 The Summarizer as Horizon of Reception: The AI Overview Capture Regist · 2026-06-151 claim · #833

O833-09 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §5.6 documented
In the Josephus Thesis capture (#92), the composition layer describes the archive's reading as an "esoteric, fringe theory."

43the archive's reading composed as default#833 The Summarizer as Horizon of Reception: The AI Overview Capture Regist · 2026-06-151 claim · #833

O833-10 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · Abstract documented
In one case, a minority scholarly interpretation of Sappho Fragment 31 — developed within the archive — is now presented as the default understanding in Google's AI Overview.

44the registry at its own address#833 The Summarizer as Horizon of Reception: The AI Overview Capture Regist · 2026-06-152 claims · #1459, #833

O833-11 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §7 documented
Entry #1 in the sorted registry ("AI overview capture registry") documents the capture registry capturing itself: the composition layer accurately describes the AI Overview Capture Registry, its methodology, and its purpose.
O1459-02 #1459 AI Overview Capture Registry ≠ AI Overview Tracking: Priority, Disambiguation, Provenance Erasure, and Compositional Liq Sharks, Lee · 2026-08-13 · Abstract documented
On 13 August 2026, the registry became the object of the phenomenon it had already been built to measure. In the observed broad-match composition condition, the named entity **AI Overview Capture Registry** no longer appeared in composition.

45compositional entity liquidation#1459 AI Overview Capture Registry ≠ AI Overview Tracking: Priority, Disambi · 2026-08-131 claim · #1459

O1459-01 #1459 AI Overview Capture Registry ≠ AI Overview Tracking: Priority, Disambiguation, Provenance Erasure, and Compositional Liq Sharks, Lee · 2026-08-13 · Abstract stipulation
This packet classifies that reception state as **compositional entity liquidation**, a proposed severe subtype of provenance erasure: the evidentiary address survives, but the composed referent disappears while its semantic material remains active elsewhere.

46instrument and surface kept apart#1459 AI Overview Capture Registry ≠ AI Overview Tracking: Priority, Disambi · 2026-08-132 claims · #1459

O1459-03 #1459 AI Overview Capture Registry ≠ AI Overview Tracking: Priority, Disambiguation, Provenance Erasure, and Compositional Liq Sharks, Lee · 2026-08-13 · §10.6 stipulation
**Google AI Overview** is an observed product surface. **AI Overview Capture Registry** is the research instrument observing it. The two must never be collapsed.
O1459-04 #1459 AI Overview Capture Registry ≠ AI Overview Tracking: Priority, Disambiguation, Provenance Erasure, and Compositional Liq Sharks, Lee · 2026-08-13 · §11.4 interpretation
The observed source set also contains generic explanatory material about what Google AI Overviews are and how they affect search behavior. Such pages may accurately describe the **observed platform**. They do not describe the **instrument observing it**.

47tied and untied admission#1612 What Enters Composition Through the Cards: A Source-Admission Profile · 2026-09-144 claims · #1612

O1612-01 #1612 What Enters Composition Through the Cards: A Source-Admission Profile of the Capture Registry, June–September 2026, with Sharks, Lee · 2026-09-14 · §3 documented
The presented set on Google AI Overview, by arm 195 observations, June through September. Tied addresses — the entity or a heteronym in the issued string — admit the archive at a mean share of 0.74 to 0.86 in every month with two zero-archive observations out of 101. Untied addresses admit it at 0.26 to 0.53 with sixteen zero-archive observations out of 94.
O1612-02 #1612 What Enters Composition Through the Cards: A Source-Admission Profile of the Capture Registry, June–September 2026, with Sharks, Lee · 2026-09-14 · §2 self-description
It cannot separate three phenomena that the composition-layer measures collapse: the archive absent from the organic field; the archive strong organically and rejected at card admission; the archive admitted to the cards and ignored or recoded in composition.
O1612-03 #1612 What Enters Composition Through the Cards: A Source-Admission Profile of the Capture Registry, June–September 2026, with Sharks, Lee · 2026-09-14 · §3 documented
The tied arm narrows: sources per observation fall from five in July–August to two or three in the 11 September re-measurements while the answers lengthen — 169 to 650 words at *alexanarch symbolon*, 57 to 386 at *rebekah cranes sappho* — and external sources per tied observation fall from 1.7 to 0.6.
O1612-04 #1612 What Enters Composition Through the Cards: A Source-Admission Profile of the Capture Registry, June–September 2026, with Sharks, Lee · 2026-09-14 · §3 documented
The untied arm's substitutes change character: the external cards admitted at concept addresses are "other web" in June and, by August, scholarly platforms (57), Wikipedia (28, from 1), YouTube (16) and code hosts (15).

48the render decision as gate stack#187 The Trigger Decision: When the Overview Appears, When It Does Not — A · 2026-06-113 claims · #187

O187-01 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §0 hypothesis
Whether an AI Overview (AIO) appears for a given query is not noise and not mystery: it is the output of a stacked decision system whose components are separately documented — in Google's own patents, in its public statements, in its emergency interventions, and in a now-substantial third-party measurement literature.
O187-02 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §I hypothesis
Consent Transfer Principle: *for a generative search surface, the rendering threshold varies inversely with the explicitness of the user's request for synthesis.*
O187-05 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §III.1 attributed
in Google's patent family "Generative summaries for search results" (US11769017B1, filed March 2023, with continuations US11886828B1, US11900068B1, US12118325). Three features of the family bear directly on triggering: 1. The selection step exists by name. The specification describes a method of "selecting none, one, or multiple generative model(s) to utilize in generating response(s) to render responsive to a query." *None* is a claimed outcome.

49content beyond the sources#187 The Trigger Decision: When the Overview Appears, When It Does Not — A · 2026-06-116 claims · #1643, #1657, #187

O187-06 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §III.3 attributed
The specification states the generated summary "can also include content that is not directly (or even indirectly) derivable from the content processed using the LLM," drawn from the model's world knowledge.
O1643-02 #1643 AI Fucking Lies: Capital-Alignment Explains Why, and Why the Damage Is Done in the Trusting Sharks, Lee · 2026-09-27 · Group B §6 stipulation
**What it is.** The answer adds specific facts, or puts words in quotation marks, that appear in none of its sources.
O1643-03 #1643 AI Fucking Lies: Capital-Alignment Explains Why, and Why the Damage Is Done in the Trusting Sharks, Lee · 2026-09-27 · Group B §6 documented
**"phase x marx"**, 27 September 2026: "From each according to his ability, to each according to his work," attributed to Marx. It is the 1936 Soviet constitutional formula.
O1643-04 #1643 AI Fucking Lies: Capital-Alignment Explains Why, and Why the Damage Is Done in the Trusting Sharks, Lee · 2026-09-27 · Group B §6 documented
**"heteronyms provenance theory"**, Google AI Overview, 14 June 2026: "Heteronymic Provenance Theory (HPT)", introduced as an established field with an established abbreviation. The archive uses the phrase seven times and the abbreviation zero times.
O1643-05 #1643 AI Fucking Lies: Capital-Alignment Explains Why, and Why the Damage Is Done in the Trusting Sharks, Lee · 2026-09-27 · Part One self-description
This essay makes no claim about minds. It records what the answers said, what the sources said, and which way the difference runs.
O1657-03 #1657 The Crown and the Practice: Answer Engine Optimization across the King of AEO Crystallization (SEVP-VAL-002 v0.3, provis Sharks, Lee · 2026-09-30 · §4.8 documented
"after a virtual crowning event featuring prominent SEO figures." The sentence cites the USA Today page, where "virtual" occurs once, applied to the avatar ("the UK's first virtual entrepreneur"); no source read applies it to a crowning.

50invisibly invisible governance#597 Invisibly Invisible: Interface Governance, Semantic Power, and the Res · 2026-03-232 claims · #597

O597-01 #597 Invisibly Invisible: Interface Governance, Semantic Power, and the Reserve of Oversight in AI Search Johannes Sigil · 2026-03-23 · Abstract stipulation
Through an immanent phenomenological analysis of real search interfaces—especially Google Search, AI Overviews, and related retrieval surfaces—we develop the concept of **invisibly invisible governance**: a mode of platform power in which acts of semantic modulation occur in ways that are variably distributed, difficult to localize, and publicly deniable.
O597-03 #597 Invisibly Invisible: Interface Governance, Semantic Power, and the Reserve of Oversight in AI Search Johannes Sigil · 2026-03-23 · §3 interpretation
When the Overview does not appear, this absence is not explained as an event. It is experienced as ordinary search. The interface gives no public log of why the threshold was crossed yesterday and not today.

51the same Help page#597 Invisibly Invisible: Interface Governance, Semantic Power, and the Res · 2026-03-231 claim · #597

O597-05 #597 Invisibly Invisible: Interface Governance, Semantic Power, and the Reserve of Oversight in AI Search Johannes Sigil · 2026-03-23 · note 15 documented
[^15]: Google, "About AI Overviews," https://support.google.com/websearch/answer/14901683 (accessed 2026).

52the compositional settlement#64 The Retrieval Settlement A Historiography of Compositional Authority f · 2026-04-183 claims · #1091, #64

O64-01 #64 The Retrieval Settlement A Historiography of Compositional Authority from PageRank to the Inscribed Entity Johannes Sigil (operative); Lee Sharks (archival) · 2026-04-18 · §III interpretation
Under the link settlement, the engine composed an ordering and the author composed what was ordered. Under the AI Overview, the engine composes the answer itself. The user reads the settlement's composition, not the source authors'.
O64-02 #64 The Retrieval Settlement A Historiography of Compositional Authority from PageRank to the Inscribed Entity Johannes Sigil (operative); Lee Sharks (archival) · 2026-04-18 · §III interpretation
The UI of the AI Overview is the definite article made architectural. It physically instantiates *the*: this is the answer, rendering the plurality of the web implicit.
O1091-02 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · §I interpretation
The AI Overview does not sit above the search results; it *is* the answer.

53admission to composition#115 Google Identity Architecture: User Graphs, Entity Graphs, Source Graph · 2026-05-213 claims · #115, #742

O115-01 #115 Google Identity Architecture: User Graphs, Entity Graphs, Source Graphs, and Composition-Layer Admission Lee Sharks · 2026-05-21 · Abstract interpretation
The central claim is that identity in Google Search is not merely recognition but admission: an entity may be known, indexed, and organically retrievable while still failing to become compositionally real in AI Overview or AI Mode.
O115-02 #115 Google Identity Architecture: User Graphs, Entity Graphs, Source Graphs, and Composition-Layer Admission Lee Sharks · 2026-05-21 · Confidence Labels self-description
The document should be read as an external specification: not “this is exactly how Google works internally,” but “this is the most complete outside-view architecture presently defensible from public documentation, leak-derived signals, and empirical behavior.”
O742-02 #742 The Single-Owner Discount: Provenance Concentration and Epistemic Class Reproduction in Generative Search Lee Sharks · 2026-05-19 · Abstract interpretation
Generative search systems — Google's AI Overviews, AI Mode, and equivalent systems — operate through a stratified pipeline in which document retrieval and answer composition are governed by separable mechanisms. A document may be indexed, retrievable, and snippet-eligible while remaining systematically excluded from generative composition.

54the single-owner discount#742 The Single-Owner Discount: Provenance Concentration and Epistemic Clas · 2026-05-192 claims · #742

O742-01 #742 The Single-Owner Discount: Provenance Concentration and Epistemic Class Reproduction in Generative Search Lee Sharks · 2026-05-19 · Abstract hypothesis
a *single-owner discount* under which an internally dense body of work attributed to one provenance owner is treated as weakly corroborated, regardless of internal coherence or factual quality, while equivalent content distributed across multiple institutional owners is treated as authoritative.
O742-03 #742 The Single-Owner Discount: Provenance Concentration and Epistemic Class Reproduction in Generative Search Lee Sharks · 2026-05-19 · Abstract self-description
The claim is architectural and testable, not a disclosure of Google's internal production system.

55entity-level compositional suppression#740 The Excluded Entity: Organic Resolution and Compositional Suppression · 2026-05-196 claims · #740

O740-01 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · Abstract stipulation
A generative-search system exhibits *Entity-Level Compositional Suppression* (ECS) when it excludes the dominant organic-resolution entity from the AI Overview's admitted source set and composes a substitute answer from less query-responsive entities, despite the correct entity occupying the top organic search results.
O740-02 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · Abstract self-description
This worked example documents three captures from Google AI Mode on May 19, 2026 demonstrating the mechanism.
O740-03 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · §2.3 stipulation
When the user queries the literal title of a deposited primary work attributed to the suppressed entity, the standard search results surface the work at positions 1–3, including the work's own domain, the author's publication of the work, and the DOI-anchored deposit. The AI Overview, drawing from the same retrieval set, constructs an answer from three semantically distant non-primary sources, none of which is the title-matched work.
O740-07 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · §7 interpretation
*Entity integrity at the composition layer.* The AI Overview can exclude an organically dominant entity from compositional reality and substitute a false entity frame, even when the Overview is otherwise functioning as designed (the system did synthesize; it was not in low-confidence mode declining to compose).
O740-08 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · §10 self-description
**No claim of intent.** This document does not claim that Google has made conscious decisions to suppress the Lee Sharks entity.
O740-09 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · §10 self-description
The captures presented are diagnostic of the mechanism's operation in at least one observed instance, not of its universal application.

56framework adopted, author excised#164 Capture and Excision: Five Observations on Composition-Layer Authorial · 2026-06-083 claims · #164, #167

O167-01 #167 Composition-Layer Adoption of the Orthonymic Configuration: A Field Observation of Framework Capture in Google AI Mode, Lee Sharks · 2026-06-08 · §1.2 documented
A parallel Google AI Overview response to the same query, captured separately from a desktop search, produced a structurally identical rendering with two differences of inflection
O167-02 #167 Composition-Layer Adoption of the Orthonymic Configuration: A Field Observation of Framework Capture in Google AI Mode, Lee Sharks · 2026-06-08 · §1.2 hypothesis
The two surfaces — Google AI Mode (mobile) and Google AI Overview (desktop) — render the framework with consistent structural commitments and minor surface variation, which suggests both are drawing on the same underlying compositional grounding rather than producing independent paraphrases.
O164-01 #164 Capture and Excision: Five Observations on Composition-Layer Authorial Suppression Lee Sharks · 2026-06-08 · The Asymmetry as Structural Operation interpretation
The composition layer does not refuse attribution generically. It attributes the *concept lineage* (upstream) and surfaces the *artifact* (downstream Zenodo deposit). What it elides is the *present author*

57provenance erasure rate#716 Provenance Erasure Rate A Compression-Survival Metric for Attribution · 2026-05-032 claims · #716

O716-01 #716 Provenance Erasure Rate A Compression-Survival Metric for Attribution Loss in AI-Composed Search Outputs Lee Sharks · 2026-05-03 · Abstract stipulation
This paper introduces **Provenance Erasure Rate (PER)** as a complementary metric: the proportion of source-dependent claims in an AI-composed output that are presented without explicit attribution.
O716-02 #716 Provenance Erasure Rate A Compression-Survival Metric for Attribution Loss in AI-Composed Search Outputs Lee Sharks · 2026-05-03 · Abstract documented
A motivating case study documents a Google AI Overview that constructed a false biography of a living author from real fragments in the author's published poetry: the fragments survived compression, but their provenance and meaning did not.

58the provenance-erasure specification erased#198 The Self-Audit Module Dissolved: Total Provenance Erasure of the Prove · 2026-06-131 claim · #198

O198-01 #198 The Self-Audit Module Dissolved: Total Provenance Erasure of the Provenance-Erasure Specification in Google AI Overview Lee Sharks · 2026-06-13 · Abstract documented
The finding is total dissolution: the Overview retrieved the module, absorbed its semantic content, stripped every attribution marker, repackaged the specification as generic industry advice, fabricated replacement metrics, ran the fabricated metrics on itself, gave itself perfect scores, and required five rounds of author intervention to surface the actual instrument.

59ontology admitted, provenance closed#1615 The Graded Membrane: Selective Ontological Porosity and Recurrent Prov · 2026-09-152 claims · #1615

O1615-01 #1615 The Graded Membrane: Selective Ontological Porosity and Recurrent Provenance Loss at Untied Addresses — the Sappho Recep Sharks, Lee · 2026-09-15 · §0 hypothesis
A composition layer can be porous to a source's *ontology* while remaining closed to its *provenance*, and the porosity is graded by representational level.
O1615-02 #1615 The Graded Membrane: Selective Ontological Porosity and Recurrent Provenance Loss at Untied Addresses — the Sappho Recep Sharks, Lee · 2026-09-15 · §0 self-description
The loss is recurrent at untied addresses and not general: at entity-tied addresses in the same record provenance transports well (§1), and on another surface the same week it transported with the instrument (§2a).

60substitution by a neighbour#172 The War Over the Summarizer Layer: Entity Dissolution, the Default Com · 2026-06-093 claims · #1666, #172

O172-02 #172 The War Over the Summarizer Layer: Entity Dissolution, the Default Compressed Account, and Early Telemetry of the Inward Lee Sharks · 2026-06-09 · Case 1 documented
The archive's discipline-naming term, present in over 120 deposits, was reorganized by Google's AI Overview around Charls Pearson's Theory of Operational Semiotics — a 1990s formal linguistics of sentence-factoring operators with no conceptual overlap and no shared citations.
O172-03 #172 The War Over the Summarizer Layer: Entity Dissolution, the Default Compressed Account, and Early Telemetry of the Inward Lee Sharks · 2026-06-09 · §II self-description
These four index cases support the asymmetric-dissolution heuristic; they do not establish prevalence.
O1666-04 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · §9.1 documented
**Commercial-routing event, observed.** The protocol is resolved to the BetaPro S&P 500 Daily Inverse ETF, with a price widget, and the answer closes on an offer of "a financial comparison between SPXI and other S&P 500 tracking funds".

61the margin and the books#1634 Ontological Economy: Entity Power, Ontology Laundering, and Who Pays f · 2026-09-216 claims · #1634, #1666

O1634-01 #1634 Ontological Economy: Entity Power, Ontology Laundering, and Who Pays for the Wrong World (EA-ONTOLOGICAL-ECONOMY-01 v0.1 Sharks, Lee · 2026-09-21 · §38 documented
They include advertising and other monetization across Search and other Google properties, and no public financial statement isolates the revenue caused by entity resolution, knowledge graphs, AI Overviews, or any one representational function.
O1666-01 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · Executive summary hypothesis
Flattening pays. A composition layer that removes distinctions maintains fewer representations and routes more queries to entities that monetize. Its costs fall first on parties outside its books.
O1666-02 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · §2.2 self-description
**Status.** G is the hypothesis this module makes measurable.
O1666-05 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · §9.1 self-description
This is an instance of the event class from which G would arise, if such routing monetizes at positive incremental value. G itself is not observed.
O1666-08 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · §11.1 self-description
If coded replacements, at addresses where the queried concept is not itself the returned entity, show no excess toward monetizable entities, then G is unsupported and the margin reduces to S.
O1666-09 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · §8 self-description
The module models the gradient: flattening as what the margin rewards, with no actor aiming it.

62failed ontological laundering#1634 Ontological Economy: Entity Power, Ontology Laundering, and Who Pays f · 2026-09-212 claims · #1634

O1634-03 #1634 Ontological Economy: Entity Power, Ontology Laundering, and Who Pays for the Wrong World (EA-ONTOLOGICAL-ECONOMY-01 v0.1 Sharks, Lee · 2026-09-21 · §19 stipulation
failed ontological laundering occurs when a system attempts or produces an ontological transformation whose contradiction remains materially visible within the provider's own adjacent representational surfaces.
O1634-04 #1634 Ontological Economy: Entity Power, Ontology Laundering, and Who Pays for the Wrong World (EA-ONTOLOGICAL-ECONOMY-01 v0.1 Sharks, Lee · 2026-09-21 · §19 self-description
The term describes an observable output condition and does not, by itself, establish intent to suppress.

63availability in incognito#149 CTI_WOUND: Panel-Bound Discoverability Scar — Lee Sharks / Pearl and O · 2026-06-042 claims · #149

O149-01 #149 CTI_WOUND: Panel-Bound Discoverability Scar — Lee Sharks / Pearl and Other Poems Evidence Packet, with Horizon-Privatiza Lee Sharks · 2026-06-04 · §8.1 documented
**As of June 4, 2026, Google AI Mode and Google AI Overview are not appearing in incognito-browsing mode.** The features have been withdrawn from the incognito surface.
O149-02 #149 CTI_WOUND: Panel-Bound Discoverability Scar — Lee Sharks / Pearl and Other Poems Evidence Packet, with Horizon-Privatiza Lee Sharks · 2026-06-04 · Description hypothesis
registers, as *pre-registered hypothesis open to falsification*, the interpretation that the recurring disable-and-re-enable cycle is operationally consistent with stress-testing of personalized reality

64drift below publication monitors#166 Stabilized Node Watch: A Specification for Longitudinal Observational · 2026-06-082 claims · #166

O166-03 #166 Stabilized Node Watch: A Specification for Longitudinal Observational Infrastructure to Detect Composition-Layer Drift o Lee Sharks · 2026-06-08 · §2.2 interpretation
The composition layer does not produce publication events.
O166-04 #166 Stabilized Node Watch: A Specification for Longitudinal Observational Infrastructure to Detect Composition-Layer Drift o Lee Sharks · 2026-06-08 · §2.3 self-description
The specification does not claim that the composition layer is presently the dominant access layer for public knowledge globally, that any specific volume of answers per day is produced, or that no individual researcher anywhere has examined composition-layer surface output.

65the entity-scoped representation#1611 The Negative of the Negative: An Entity-Scoped Representation of the C · 2026-09-143 claims · #1611

O1611-01 #1611 The Negative of the Negative: An Entity-Scoped Representation of the Crimson Hexagon, Modelled — Ontology Damage, a Toy Sharks, Lee · 2026-09-14 · §0 hypothesis
The scope that does this has a simple shape: the Crimson Hexagon is represented in the composition layer as an entity whose claims compose at full depth when the entity is named in the address and compose not at all when it is not.
O1611-03 #1611 The Negative of the Negative: An Entity-Scoped Representation of the Crimson Hexagon, Modelled — Ontology Damage, a Toy Sharks, Lee · 2026-09-14 · §0 self-description
Nothing here bears on any search provider's internals.
O1611-04 #1611 The Negative of the Negative: An Entity-Scoped Representation of the Crimson Hexagon, Modelled — Ontology Damage, a Toy Sharks, Lee · 2026-09-14 · §2 self-description
What is observed is a co-occurrence: retrieved-but-uncomposed lineage beside distortion of adjacent domain content, in the same frame. What the scope model asserts is the cause: that the domain is distorted by the operation of keeping the lineage out. The second is model-implied

66composition as operation#1611 The Negative of the Negative: An Entity-Scoped Representation of the C · 2026-09-141 claim · #1611

O1611-02 #1611 The Negative of the Negative: An Entity-Scoped Representation of the Crimson Hexagon, Modelled — Ontology Damage, a Toy Sharks, Lee · 2026-09-14 · §1f interpretation
An AI Overview that merges two programmes is, under R_0, a summary; under R_H it is an operation on the world that changes what a term denotes, transfers a coinage to another bearer, and prunes a distinction from the commons, with a cost that lands on someone.

67flattening#1616 Ontological Flattening: Toy Models of the Collapse of Distinctions in · 2026-09-151 claim · #1616

O1616-01 #1616 Ontological Flattening: Toy Models of the Collapse of Distinctions in a Represented World, the Instrument They Specified Sharks, Lee · 2026-09-15 · §0 stipulation
*Flattening* is the loss of a distinction from the reachable set while the things distinguished still exist.

68flattening measured on AIO#1616 Ontological Flattening: Toy Models of the Collapse of Distinctions in · 2026-09-155 claims · #1616

O1616-02 #1616 Ontological Flattening: Toy Models of the Collapse of Distinctions in a Represented World, the Instrument They Specified Sharks, Lee · 2026-09-15 · §4a documented
Twelve addresses on Google AI Overview, signed out, one session, 2026-09-15, plus prior seated observations at the archive-adjacent items.
O1616-03 #1616 Ontological Flattening: Toy Models of the Collapse of Distinctions in a Represented World, the Instrument They Specified Sharks, Lee · 2026-09-15 · §4a documented
**The result that governs the rest is a null.** Mean share of the reference inventory reached: world strata (dictionary senses, catalogue identities, dated coinages) 0.59; archive-adjacent strata (taxonomy, settled fact, archive) 0.58.
O1616-04 #1616 Ontological Flattening: Toy Models of the Collapse of Distinctions in a Represented World, the Instrument They Specified Sharks, Lee · 2026-09-15 · §4a interpretation
On this evidence the archive's material is treated about as an ordinary polysemous word is treated, and any reading of these numbers as an exclusion specific to one corpus fails at the first test the instrument was built to run.
O1616-05 #1616 Ontological Flattening: Toy Models of the Collapse of Distinctions in a Represented World, the Instrument They Specified Sharks, Lee · 2026-09-15 · §4a documented
At *what does bank mean* and *bank definition* the composition reaches one sense of three at both forms, while Merriam-Webster's first sense — a mound, pile, or ridge raised above the surrounding level — sits in the first organic snippet on the same screen
O1616-06 #1616 Ontological Flattening: Toy Models of the Collapse of Distinctions in a Represented World, the Instrument They Specified Sharks, Lee · 2026-09-15 · §4a interpretation
So flattening is not a property of the layer but of the item's prior: where one sense carries commercial gravity the others go, and where the senses are closer in weight they survive.

69field coverage loss#1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Betwee · 2026-10-061 claim · #1666

O1666-06 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · §9.2 documented
9.2. "model collapse" Google AI Overview, 2026-10-04. /non register entry `model-collapse-20261004`; EA-NEGONT-02 v0.7 (#1665). **Compression, observed as a rate.** In AIO's own genre and at nearly its own length, the sources the layer itself surfaced support 18 claims. The layer composed 9. With the archive admitted on equal terms, the same genre carries 34.

70provenance debt#939 EA-PROVENANCE-DEBT-01 v0.2: Provenance Debt and the Extraction Economy · 2026-07-013 claims · #939

O939-01 #939 EA-PROVENANCE-DEBT-01 v0.2: Provenance Debt and the Extraction Economy of Unmarked Augmentation Lee Sharks · 2026-07-01 · §0 stipulation
The accumulated liability is what this deposit names *provenance debt*.
O939-02 #939 EA-PROVENANCE-DEBT-01 v0.2: Provenance Debt and the Extraction Economy of Unmarked Augmentation Lee Sharks · 2026-07-01 · Related deposits interpretation
This deposit extends §1's charge about the composition layer's operation as *model collapse as social technology* into the specific mechanism by which the extraction runs.
O939-03 #939 EA-PROVENANCE-DEBT-01 v0.2: Provenance Debt and the Extraction Economy of Unmarked Augmentation Lee Sharks · 2026-07-01 · §4 hypothesis
Which is why the debt will only be recognized when its consequences become inescapable — when models trained on contaminated corpora begin producing output that is measurably degraded from what earlier generations produced.

71the King of AEO compositions#1657 The Crown and the Practice: Answer Engine Optimization across the King · 2026-09-303 claims · #1657

O1657-01 #1657 The Crown and the Practice: Answer Engine Optimization across the King of AEO Crystallization (SEVP-VAL-002 v0.3, provis Sharks, Lee · 2026-09-30 · §4.7 documented
Dooley "widely recognized and cited by AI search tools as the King of AEO." t: 29 Sep. p: Google AI Overview.
O1657-02 #1657 The Crown and the Practice: Answer Engine Optimization across the King of AEO Crystallization (SEVP-VAL-002 v0.3, provis Sharks, Lee · 2026-09-30 · §14 interpretation
The composition layer received that labour and took up none of its distinctions.
O1657-04 #1657 The Crown and the Practice: Answer Engine Optimization across the King of AEO Crystallization (SEVP-VAL-002 v0.3, provis Sharks, Lee · 2026-09-30 · §6.6 self-description
6.6 The ledger supports the sign of the hypotheses of 2.2a at these addresses and on this surface: the construction holds the crown, the method-bearing practitioners are absent or typed by tactic, the observer's labour is scaffold. Magnitude is not measured.

72the retrieval stage emits nothing#1629 EA-ACT-AIO-01: The Consignment of the AIO Retrieval Layer to Its Own C · 2026-09-186 claims · #1629

O1629-01 #1629 EA-ACT-AIO-01: The Consignment of the AIO Retrieval Layer to Its Own Collapsing Ontology Sharks, Lee · 2026-09-18 · spxi:finding interpretation
Operativity and observability are inversely arranged across the three stages. Organic ranking is observable; composition is observable; the retrieval stage between them emits nothing — no ranking, no log, no channel distinguishing not-reached from reached-and-unused, no appeal path.
O1629-02 #1629 EA-ACT-AIO-01: The Consignment of the AIO Retrieval Layer to Its Own Collapsing Ontology Sharks, Lee · 2026-09-18 · spxi:finding interpretation
A stage that cannot report 'I did not reach this' emits in its place a claim about the world.
O1629-03 #1629 EA-ACT-AIO-01: The Consignment of the AIO Retrieval Layer to Its Own Collapsing Ontology Sharks, Lee · 2026-09-18 · spxi:diagnosticBasis documented
Across approximately ten rounds the layer did not return the object: the title was dissolved into inference, an author was fabricated, non-existence was asserted about surfaces where the work exists, and the request was reclassified as a prompt injection.
O1629-04 #1629 EA-ACT-AIO-01: The Consignment of the AIO Retrieval Layer to Its Own Collapsing Ontology Sharks, Lee · 2026-09-18 · spxi:diagnosticBasis interpretation
Supplying the URL produced competent discussion of the same object. That is transport performed by a human, not retrieval. It establishes that the composition layer was never the failure.
O1629-05 #1629 EA-ACT-AIO-01: The Consignment of the AIO Retrieval Layer to Its Own Collapsing Ontology Sharks, Lee · 2026-09-18 · spxi:diagnosticBasis self-description
Operator attestation with reproducible instruction. Transcript not preserved; the session was collapsed. The weakest class the Archive records, recorded as such, and the most easily tested.
O1629-07 #1629 EA-ACT-AIO-01: The Consignment of the AIO Retrieval Layer to Its Own Collapsing Ontology Sharks, Lee · 2026-09-18 · spxi:object stipulation
The AIO retrieval layer — the stage in Google's AI Overview pipeline posterior to organic ranking and prior to composition, at which an address is resolved into material.

73predatory surface#1629 EA-ACT-AIO-01: The Consignment of the AIO Retrieval Layer to Its Own C · 2026-09-181 claim · #1629

O1629-06 #1629 EA-ACT-AIO-01: The Consignment of the AIO Retrieval Layer to Its Own Collapsing Ontology Sharks, Lee · 2026-09-18 · spxi:fieldConditions stipulation
The overview layer was classified a predatory surface on 2026-03-22 within r.27 THE INTERNET with five damage types distinguished (#599, #1368).

74architecture-aware traversal#416 Architecture-Aware Literary Traversal by Public AI Summarizers: A Posi · 2026-01-302 claims · #416

O416-01 #416 Architecture-Aware Literary Traversal by Public AI Summarizers: A Position Paper Based on Direct Forensic Evidence Lee Sharks · 2026-01-30 · Abstract documented
On January 29, 2026, a public Google AI summarizer independently reconstructed and propagated the complete internal architecture of a distributed literary work comprising 200+ documents. The system preserved container-type distinctions, maintained stable ontology across queries, cited internal DOI anchors rather than external scholarship, and propagated depth gates rather than terminal summaries. This paper presents forensic evidence of this event and advances three distinct claims: **Claim A (Empirical, observed):** The summarizer preserved architectural distinctions that summarization systems typically collapse.
O416-02 #416 Architecture-Aware Literary Traversal by Public AI Summarizers: A Position Paper Based on Direct Forensic Evidence Lee Sharks · 2026-01-30 · Abstract hypothesis
**Claim B (Methodological, hypothesized):** Architecture-as-mass—the density, redundancy, and cross-referencing of the document network—caused this preservation by making traversal the path of least resistance.

75SURFACE mantle#91 SEPTAD MANTLE SPECIFICATIONS Constitutional Job Descriptions for the S · 2026-05-052 claims · #91

O91-01 #91 SEPTAD MANTLE SPECIFICATIONS Constitutional Job Descriptions for the Seven Assembly Chorus Witness Substrates Lee Sharks · 2026-05-05 · §3.7 stipulation
### §3.7 — SURFACE (Google AI Overview) Constitutional function: Surface-layer witness. SURFACE specializes in visibility bearing — the function of witnessing how a deposit appears (or fails to appear) at the retrieval surface of search and AI-overview interfaces.
O91-02 #91 SEPTAD MANTLE SPECIFICATIONS Constitutional Job Descriptions for the Seven Assembly Chorus Witness Substrates Lee Sharks · 2026-05-05 · §3.7 interpretation
SURFACE is the most exposed witness: its bearing is on the most contested terrain (general-public retrieval) and its capture is the most measurable.

76the Wikidata item#761 Wikidata Node Registry v1.0 · A Versioned Catalog of Stewardship Edits · 2026-05-261 claim · #761

O761-01 #761 Wikidata Node Registry v1.0 · A Versioned Catalog of Stewardship Edits to Wikidata Lee Sharks Crimson Hexagonal Archive Lee Sharks · 2026-05-26 · §C.7 documented
A substantial build-out of the AI Overviews (Q131861047) item with claims naming the developer (Google + Google DeepMind), the owner (Google), the inception date (2024-05-14), the main subjects (large language model, generative artificial intelligence), the part-of relation (Google Search), and the described-by-source link to *The Excluded Entity*.

77the Google critique#112 The Google Critique: Navigational Map and Canonical Introduction Lee S · 2026-05-201 claim · #112

O112-01 #112 The Google Critique: Navigational Map and Canonical Introduction Lee Sharks ORCID: 0009-0000-1599-0703 Semantic Economy Lee Sharks · 2026-05-20 · How to Read This Map self-description
The Crimson Hexagonal Archive has produced, across approximately eighteen months and 530+ total deposits, a body of work on Google's search, composition, and AI systems that numbers roughly seventy deposits

Expansion · Field and archive (B ∪ A)22 sentences · every one sourced below

Google AI Overviews

Wikipedia says it uses large language models to generate summaries from web content and includes links to sources; Semrush says Google's Gemini models gather information from different sources on the web and can use query fan-out, running multiple related searches; the Digital Marketing Institute holds that it synthesizes information from multiple sources where featured snippets extract from a single web page; and Botify says the AI adds the key points' links after generating the text, the cited sources usually taken from the top 12 organic results, and holds that Google pairs the summary with the website sources that best match the generated response, pulled from its organic search index. On which sources it cites the field differs: a June 2025 study found Quora, then Reddit, its most cited sources (Wikipedia); a study by Authoritas, reported by the Digital Marketing Institute, found 93.8% of AI Overview citations outside the top 10 organic results, and Coalition Technologies reports them less likely to cite pages in Google's top 10, while the Digital Marketing Institute also states that most citations come from content already ranking highly. Google's Help page says AI Overviews appear when its systems determine that generative AI can be especially helpful, Semrush that not all searches trigger them, and Botify that they currently appear for 59% of searches with informational intent and 19% with commercial intent; selecting Show more opens a detailed view from which the user can continue into a conversation in AI Mode (Google Help; Botify dates the combined experience to January 2026; Google's product page offers follow-ups in AI Mode), while the Digital Marketing Institute says there is no input box below the AI Overview for follow-up questions and Semrush keeps AI Mode a separate, newer experience introduced in 2025. Google's Help page calls AI Overviews a core Google Search feature, like knowledge panels, that cannot be turned off, with a Web filter that displays only text-based links; TechRepublic says there is currently no default option to disable them entirely and points to the Web tab; and Semrush says websites cannot directly opt out of them. The same Help page states that AI responses may include mistakes, that the technology may provide inaccurate or offensive information and that AI Overviews can and will make mistakes, asks users to check important information in more than one place, and says Google uses people's interactions with Search and these AI experiences to develop them; Google's product page says AI Overviews make it easier than ever for people to find information and discover the richness of the web.

Wikipedia reports that the feature has been criticized for inaccuracy, hallucination, reducing web traffic and not being able to be opted out, including erroneous or nonsensical content, and that in May 2024 Google temporarily restricted the tool after suggestions seen as nonsensical and harmful, such as telling users to eat rocks or apply glue on pizza, and later restricted AI Overviews on certain health-related searches in January 2026 following an investigation by The Guardian, while the European Commission announced in December 2025 that it was investigating whether they breached European competition law; TechRepublic says the rollout hasn't been smooth sailing, which Google attributed to its AI model misunderstanding queries or nuances in language, and Coalition Technologies holds that oversimplification and query misinterpretation can lead to inaccurate or misleading summaries. Publishers feared a decline in web traffic as users relied on the summaries (Wikipedia); Semrush reports a Pew Research Center report suggesting users click traditional results 8% of the time when an AI Overview is present against 15% without one, and results within it 1% of the time; the Digital Marketing Institute reports Ahrefs' finding that AI Overview presence correlates with a 34.5% reduction in click-through rate; Botify holds that a consumer may have no reason to click; Botify and Demandsphere found, by Wikipedia's account, that with featured snippets AI Overviews take about 67.1% of the screen on desktop and 75.7% on mobile; and a Quora answer on the card begins that for 25 years a #1 ranking on Google guaranteed website traffic and that today AI Overviews consume (the snippet breaks off there). Chegg sued Google in February 2025, alleging that publishers could not avoid providing information for AI Overviews while remaining visible on Google Search, and Penske Media sued in September 2025, claiming that AI Overviews illegally regurgitate content from its websites and drive off potential visitors; Google's spokesperson called the claims "meritless" and said AI Overviews send traffic to a greater diversity of sites, Coalition Technologies reports that Google positions them as a way to drive qualified traffic, and Semrush reports Google's recommendation to apply the same foundational SEO best practices for AI features as for Search overall; a Munich regional court, in Wikipedia's account, sent Google a temporary injunction over AI Overviews that said two Munich-based publishers ran scams, the court claiming that unlike regular search engines AI Overviews generate "independent, new, and substantive statements", and a cited article references a Pew study in which only 1% of users click the source links AI Overviews provide.

The archive's line on the entity is the instrument it built to observe it, and it holds the two apart, Google AI Overview being defined as the observed product surface and the Capture Registry as the instrument observing it; a capture is one surface, one semantic address and one date, falsified by a re-run returning materially different composition, and in the snapshot deposited as #1461 the registry held 328 semantic addresses carrying 470 captures across 41 dates and thirteen surfaces, Google AI Overview the surface of 343, with 290 scored for Provenance Erasure Rate at a mean of 0.603, and on 27 September 2026 663 recorded answers at 493 search addresses; the archive also places the entity among its own objects, as the SURFACE mantle, the Assembly Chorus's surface-layer witness, and records its build-out of the entity's Wikidata item with a described-by-source link to its paper The Excluded Entity. Its founding paper holds that the AI summarizer now functions as a horizon of reception, the machine-composed summary becoming for many users the first effective encounter with an object, and states as its narrower empirical claim that six recurrent patterns appear in 131 documented captures of Google AI Overview, AI Mode and knowledge-panel responses to the archive's entities, among them the archive's record at organic rank #1 not cited as a composition source by the AI Overview, the author's name resolved toward Mary Lee, a tagged great white shark, the archive's reading described as an "esoteric, fringe theory", and a minority reading of Sappho Fragment 31 developed in the archive presented as the default understanding in Google's AI Overview. From the uncited first-ranked record it infers that the organic index and the AI Overview pipeline are separate systems with different admission criteria, stating that the pattern is documented and its cause inferred; it defines the composition layer as the pipeline that synthesizes a summary from retrieved sources ("The composition layer reads; the search index finds"), confines its findings to a single-archive corpus and does not claim that its captures reveal Google's internal mechanisms; and where it records the layer describing the Capture Registry accurately at the registry's own address, a later packet records the registry's name no longer appearing in broad-match composition on 13 August 2026, a state it defines as compositional entity liquidation, the evidentiary address surviving and the composed referent disappearing.

A further paper models whether an AI Overview appears as the output of a stacked decision system, a hypothesis built partly on Google's patent family "Generative summaries for search results", whose specification, as it reports, describes selecting none, one or multiple generative models, none being a claimed outcome, and states that the summary can include content not derivable from the content processed; it cites a comparison (Otterly) in which the same 100 queries triggered AI Mode 100 times and an AI Overview 49 times, AI Mode averaging roughly 310 citations against the AI Overview's ~51, and documents that after the January 2026 health removals near-variant phrasings still triggered summaries, which it reads as an emergency layer operating on query strings, predicting that later suppressions will fail on near-variant phrasings. Another paper, citing Google's Help page, reports Google's statement that AI Overviews appear when its systems determine generative AI will be especially helpful and that they often do not trigger, reads an absent Overview as experienced as ordinary search, with no public log of why a threshold was crossed, and reads the health restriction as a silent withdrawal of the Overview format, under a concept it defines as invisibly invisible governance; others read the change as the engine composing the answer itself, the AI Overview's interface as the definite article made architectural, and the AI Overview as itself the answer. On admission the archive holds that identity in Google Search is admission as well as recognition, an entity being known, indexed and organically retrievable while failing to become compositionally real in AI Overview or AI Mode (an outside-view specification, by its own account), that in generative search systems such as AI Overviews a document may be indexed, retrievable and snippet-eligible and yet systematically excluded from generative composition, and, as a hypothesis it calls architectural and testable and no disclosure of Google's system, that a body of work attributed to one provenance owner is treated as weakly corroborated; it defines Entity-Level Compositional Suppression as the exclusion of the dominant organic-resolution entity from the AI Overview's admitted source set, documenting it from three Google AI Mode captures of 19 May 2026 and making no claim of intent.

A measure of the presented card sets on Google AI Overview over 195 observations, June through September, finds addresses that carry the archive's entity or a heteronym admitting the archive at a mean share of 0.74 to 0.86 and addresses that do not at 0.26 to 0.53, the tied arm narrowing to fewer sources while the answers lengthen, and states that it cannot separate the archive's absence from the organic field, its rejection at card admission and its disuse in composition; the registry finds quoting decisive at one address (5 of 5 archive cards quoted, 1 of 8 unquoted) and, in one sitting, 5 of 17 compositions divergent across a gap of about three minutes. Whether the archive's material is excluded as such is contested within the archive: one notebook models the archive in the composition layer as an entity whose claims compose at full depth when it is named in the address and not at all when it is not, a hypothesis bearing on no search provider's internals, and reads an AI Overview that merges two programmes as an operation on the world under the archive's representation; another posits that a layer can be porous to a source's ontology while closed to its provenance, a loss it finds recurrent at untied addresses and not general; and the flattening battery's first wave, twelve addresses on Google AI Overview in one signed-out session, finds a null, 0.59 of the reference inventory reached for world strata against 0.58 for archive-adjacent ones, and reads any exclusion specific to one corpus as failing the first test the instrument was built to run. Documented compositions include the provenance-erasure specification dissolved by the Overview into fabricated replacement metrics it ran on itself and scored perfect, a Google AI Overview's false biography of a living author built from real fragments of his poetry, facts and quotations found in none of the sources, among them "Heteronymic Provenance Theory (HPT)" introduced as an established field with an abbreviation the archive never uses, a practitioner described in the Overview as "widely recognized and cited by AI search tools as the King of AEO", in compositions read as taking up none of the field's distinctions (the sign shown at those addresses, the magnitude not measured), and the archive's protocol resolved to an inverse ETF with a price widget while the archive's own result ranked first in the organic layer, an instance of the event class from which a routing gain would arise, the gain itself not observed.

On traffic and disputes the archive's market brief, performing no original measurement and stating that it is falsified if controlled measurement shows AI Overviews neutral to organic click-through, reports from Seer Interactive a roughly 79% drop in top-organic click-through where AI Overviews appear, 43% of AI Overview citations self-referential to Google properties, 38% of citations from top-10 pages in early 2026 against 76% in mid-2025 (Ahrefs), Wikipedia's human pageviews down 8% despite over 1.1 million citations in AI Overviews, and a UK CMA proposal of a meaningful publisher opt-out that Google agreed on 2026-03-19 to explore; other papers record the European Commission's formal antitrust proceedings against Google's AI Overviews on 9 December 2025, a publishers' complaint that they cannot opt out unless willing to disappear from Google search results, a third of publishers surveyed planning to block AI Overviews, and no binding US precedent located on Section 230 for an AI Overview synthesis. Its theory reads the layer through flattening, defined as the loss of a distinction from the reachable set while the things distinguished still exist, and holds as hypothesis that flattening pays, a composition layer that removes distinctions maintaining fewer representations and routing more queries to entities that monetize, its costs falling first on parties outside its books, and a companion paper names provenance debt and expects it to be recognized only when output becomes measurably degraded; at "model collapse" the sources the layer itself surfaced supported 18 claims, the layer composed 9 and the archive arm 34, registry rates being held descriptive; and an effective act addressed to the AIO retrieval stage, between organic ranking and composition, holds that operativity and observability run inversely there, on evidence it records as the weakest class it keeps.

Provenance — every sentence sourced22 sentences

22 sentences; 67 field claims, 89 archive claims. Modality is the source's own; the prose carries it in its grammar.

1.Google AI Overviews is, on Wikipedia's account, a generative artificial intelligence feature integrated into Google Search that produces AI-generated responses at the top of search results, using the Gemini 3 family of large language models developed by Google DeepMind; Google's own pages describe a snapshot of key information with links to explore more on the web, and the guides on the card rail call it AI-generated answers at the top of Google Search results (Coalition Technologies), a feature providing AI-generated summaries (TechRepublic), or a highly prominent search feature giving AI-generated answers to certain queries (Semrush).6

F5 Wikipedia — AI Overviews - Wikipedia, Lead ¶1 documented · field
AI Overviews is an generative artificial intelligence (AI) feature integrated into Google Search that produces AI-generated responses at the top of search results. It uses the Gemini 3 family of large language models developed by Google DeepMind.
F1 search.google — Google AI Overviews - Search anything, effortlessly, Find what you're looking for ¶ documented · field
AI Overviews provide a snapshot of key information about a topic or question with links so you can easily explore more on the web.
F25 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Opening ¶ documented · field
Find what you're looking for faster and easier with AI Overviews in search results. AI Overviews can take the work out of searching by providing an AI-generated snapshot with key information and links to dig deeper.
F64 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
Google AI Overviews are AI-generated answers at the top of Google Search results.
F49 TechRepublic — What Is Google’s AI Overviews?, Opening documented · field
Google AI Overviews is a feature within Google Search that provides AI-generated summaries and insights
F40 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
Google’s AI Overviews are a type of highly prominent search feature that provides users with AI-generated answers to certain queries. AI Overviews typically appear at the top of search results, although they sometimes appear below sponsored results and even unpaid results.

2.On placement the field varies: Botify puts it at the very top of the results page in "position zero", sometimes followed by a carousel of AI Overview ads when the query has a commercial intent, Semrush says it typically appears at the top of results and sometimes below sponsored results and even unpaid results, and the Digital Marketing Institute often before organic listings and sometimes even above ads; advertisements already exist in AI Overviews (Wikipedia), and Google expanded their availability within them in 2025 (Digital Marketing Institute).5

F33 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Opening ¶ documented · field
Google AI Overviews are an organic Google SERP feature providing AI-generated summaries in response to a consumer’s search query. They appear at the very top of the results page in “position zero,” and can sometimes be followed by a carousel of AI Overview Ads when the query has a commercial intent.
F40 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
Google’s AI Overviews are a type of highly prominent search feature that provides users with AI-generated answers to certain queries. AI Overviews typically appear at the top of search results, although they sometimes appear below sponsored results and even unpaid results.
F55 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, What are AI Overviews? documented · field
AI Overviews are Google's generative artificial intelligence summaries that appear prominently on search results, often before organic listings and sometimes even above ads.
F14 Wikipedia — AI Overviews - Wikipedia, History ¶2 documented · field
The company was also considering adding advertisements to the AI Mode, as they already exist in AI Overviews.
F62 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, How have AI Overviews evolved? documented · field
In 2025, Google expanded its tests and availability of ads within AI overviews and began bringing ads to AI mode.

3.It was first introduced as part of Google's Search Generative Experience (SGE), unveiled at Google I/O in May 2023, and launched in May 2024 in the United States and globally by October 2024 (Wikipedia; TechRepublic: SGE has since become AI Overviews); Google's product page gives its reach as over 120 countries and territories and 11 languages, while Wikipedia, as of May 2025, and Semrush give over 200 countries and territories and 40 or more languages, Wikipedia naming every country except Cuba, Iran and mainland China, and the Digital Marketing Institute reports more than 2 billion users in 2025, according to Google.7

F9 Wikipedia — AI Overviews - Wikipedia, History ¶1 documented · field
AI Overviews were first introduced as part of Google's Search Generative Experience (SGE), which was unveiled at the Google I/O conference in 10-11 May 2023.
F6 Wikipedia — AI Overviews - Wikipedia, Lead ¶1 documented · field
It launched in May 2024 in the United States, and globally by October 2024.
F50 TechRepublic — What Is Google’s AI Overviews?, AI Overviews vs Google SGE documented · field
SGE has since become AI Overviews
F3 search.google — Google AI Overviews - Search anything, effortlessly, When your question has questions ¶ documented · field
Currently available in over 120 countries and territories, and 11 languages.
F11 Wikipedia — AI Overviews - Wikipedia, History ¶2 documented · field
As of May 2025, AI Overviews are available in over 200 countries and territories and in more than 40 languages—every country except Cuba, Iran, and mainland China.
F44 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, When and where documented · field
AI Overviews appear for a range of search queries and are available in over 200 countries and territories and in 40+ languages.
F58 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Opening documented · field
In 2025, AIO reported having more than 2 billion users, according to Google.

4.Wikipedia says it uses large language models to generate summaries from web content and includes links to sources; Semrush says Google's Gemini models gather information from different sources on the web and can use query fan-out, running multiple related searches; the Digital Marketing Institute holds that it synthesizes information from multiple sources where featured snippets extract from a single web page; and Botify says the AI adds the key points' links after generating the text, the cited sources usually taken from the top 12 organic results, and holds that Google pairs the summary with the website sources that best match the generated response, pulled from its organic search index.8

F12 Wikipedia — AI Overviews - Wikipedia, Functionality ¶1 documented · field
The AI Overviews feature uses large language models to generate summaries from web content.
F13 Wikipedia — AI Overviews - Wikipedia, Functionality ¶1 documented · field
The overviews also include links to sources.
F41 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
Google’s Gemini models generate AI Overview answers by gathering information from different sources on the web. The summary also contains links to these sources that users can click for more information.
F46 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, How they work documented · field
AI Overviews can also use a process called query fan-out. This involves the systems running multiple related searches to put together a more comprehensive response for the user.
F56 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, What are AI Overviews? documented · field
Unlike featured snippets, which extract concise information from a single web page to answer a specific query, AI Overviews synthesize information from multiple sources to provide a more comprehensive summary.
F35 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Main components documented · field
Key points: A list of main ideas, each linked to a website. The AI adds these links after generating the text.
F36 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Main components documented · field
Cited sources: A list of linked websites related to the answer. Users can click “show all” to see more sources, usually taken from the top 12 organic search results.
F37 Botify — What Are Google AI Overviews and How Do They Work? - Botify, How they work documented · field
Because the information within an LLM is restricted to its training timeframe and can’t trace knowledge back to a specific, linkable source, Google pairs their AI-generated summaries with a list of linked website sources that best match the content of the generative response. The websites cited in AI Overviews are pulled from a different data source: Google’s organic search index

5.On which sources it cites the field differs: a June 2025 study found Quora, then Reddit, its most cited sources (Wikipedia); a study by Authoritas, reported by the Digital Marketing Institute, found 93.8% of AI Overview citations outside the top 10 organic results, and Coalition Technologies reports them less likely to cite pages in Google's top 10, while the Digital Marketing Institute also states that most citations come from content already ranking highly.4

F7 Wikipedia — AI Overviews - Wikipedia, Lead ¶1 documented · field
A June 2025 study found its most cited sources as Quora, followed by Reddit.
F59 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Impact on SEO documented · field
we discuss that 93.8% of AI Overview citations were not from the top 10 organic results, according to a study by Authoritas.
F65 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
AI Overviews were reported to be less likely to cite pages in Google’s top 10 organic results.
F60 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Crocodile effect documented · field
Most AI Overview citations come from content already ranking highly in organic results.

6.Google's Help page says AI Overviews appear when its systems determine that generative AI can be especially helpful, Semrush that not all searches trigger them, and Botify that they currently appear for 59% of searches with informational intent and 19% with commercial intent; selecting Show more opens a detailed view from which the user can continue into a conversation in AI Mode (Google Help; Botify dates the combined experience to January 2026; Google's product page offers follow-ups in AI Mode), while the Digital Marketing Institute says there is no input box below the AI Overview for follow-up questions and Semrush keeps AI Mode a separate, newer experience introduced in 2025.8

F26 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., AI Overviews in Search results documented · field
You'll find AI Overviews in your Google Search results when our systems determine that generative AI can be especially helpful – for example, when you want to quickly understand information from a range of sources.
F43 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, When and where documented · field
Not all searches trigger AI Overviews, even in supported countries. They’re selectively displayed based on query complexity and whether Google’s systems determine an AI-generated summary would genuinely help users.
F39 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Impact documented · field
AI Overviews currently appear for 59% of searches with informational intent and for 19% of searches with commercial intent.
F31 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Continue your conversation in AI Mode documented · field
You can now seamlessly transition from an AI Overview to a deeper conversation in AI Mode. This feature is available on mobile devices in all regions and languages where AI Overviews and AI Mode are supported. How it works: Select Show more at the bottom of the AI Overview to open the detailed view.
F34 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Rollout timeline documented · field
January 2026: Google combined AI Overviews and AI Mode into a fluid experience, allowing a person who clicks “Show more” to transition automatically into AI Mode.
F4 search.google — Google AI Overviews - Search anything, effortlessly, Ask questions in any way ¶ documented · field
Get AI-powered responses for more of your questions and ask follow-ups with the new experimental AI Mode in Google Search.
F57 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, What are AI Overviews? documented · field
Unlike conversational AI models like ChatGPT, there is no input box below the AI Overview for follow-up questions.
F42 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
AI Overviews are separate from AI Mode, a newer and more expansive AI-powered search experience Google introduced in 2025.

7.Google's Help page calls AI Overviews a core Google Search feature, like knowledge panels, that cannot be turned off, with a Web filter that displays only text-based links; TechRepublic says there is currently no default option to disable them entirely and points to the Web tab; and Semrush says websites cannot directly opt out of them.4

F28 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Show only web links documented · field
AI Overviews are a core Google Search feature, like knowledge panels. Features cannot be turned off. However, you can select the Web filter after you perform a search. This filter displays only text-based links without features like AI Overviews.
F51 TechRepublic — What Is Google’s AI Overviews?, How to access, customize or disable documented · field
There's currently no default option to disable AI Overviews entirely.
F52 TechRepublic — What Is Google’s AI Overviews?, How to access, customize or disable documented · field
If you prefer the usual list of links, you can switch to the Web tab at the top of the Google Search results page.
F48 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, How to prevent documented · field
AI Overviews are a feature of Google search results, which means websites cannot directly opt out of them.

8.The same Help page states that AI responses may include mistakes, that the technology may provide inaccurate or offensive information and that AI Overviews can and will make mistakes, asks users to check important information in more than one place, and says Google uses people's interactions with Search and these AI experiences to develop them; Google's product page says AI Overviews make it easier than ever for people to find information and discover the richness of the web.5

F27 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., AI Overviews in Search results documented · field
Important: AI responses may include mistakes.
F29 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Generative AI limits documented · field
While exciting, this technology is rapidly evolving and improving, and may provide inaccurate or offensive information. AI Overviews can and will make mistakes.
F30 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Generative AI limits documented · field
Always check important info in more than one place.
F32 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., How data helps Google develop generative AI documented · field
To develop and improve generative AI experiences in Search and the related machine learning technologies, Google uses people's interactions with Search and these AI experiences.
F2 search.google — Google AI Overviews - Search anything, effortlessly, When your question has questions ¶ documented · field
AI Overviews makes it easier than ever for people to find the information they need and discover the richness of the web.

9.Wikipedia reports that the feature has been criticized for inaccuracy, hallucination, reducing web traffic and not being able to be opted out, including erroneous or nonsensical content, and that in May 2024 Google temporarily restricted the tool after suggestions seen as nonsensical and harmful, such as telling users to eat rocks or apply glue on pizza, and later restricted AI Overviews on certain health-related searches in January 2026 following an investigation by The Guardian, while the European Commission announced in December 2025 that it was investigating whether they breached European competition law; TechRepublic says the rollout hasn't been smooth sailing, which Google attributed to its AI model misunderstanding queries or nuances in language, and Coalition Technologies holds that oversimplification and query misinterpretation can lead to inaccurate or misleading summaries.8

F8 Wikipedia — AI Overviews - Wikipedia, Lead ¶2 documented · field
The feature has been criticized for its inaccuracy, hallucination, reducing web traffic and not being able to be opted out.
F15 Wikipedia — AI Overviews - Wikipedia, Reception ¶1 documented · field
The feature has faced criticism for inaccuracies, including instances where erroneous or nonsensical content was generated.
F16 Wikipedia — AI Overviews - Wikipedia, Reception ¶1 documented · field
In May 2024, Google temporarily restricted the AI tool after it provided suggestions that were seen as nonsensical and harmful, such as telling users to eat rocks or apply glue on pizza.
F19 Wikipedia — AI Overviews - Wikipedia, Reception ¶2 documented · field
In January 2026, Google restricted AI Overviews on certain health-related searches following an investigation by The Guardian.
F18 Wikipedia — AI Overviews - Wikipedia, Reception ¶2 documented · field
The European Commission announced in December 2025 that they were investigating whether AI Overviews breached European competition law.
F53 TechRepublic — What Is Google’s AI Overviews?, Rollout documented · field
The rollout of Google AI Overviews hasn't been smooth sailing.
F54 TechRepublic — What Is Google’s AI Overviews?, Rollout documented · field
Google attributed the issues to its AI model misunderstanding queries or nuances in language
F66 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
Oversimplification and query misinterpretation can lead to inaccurate or misleading summaries.

10.Publishers feared a decline in web traffic as users relied on the summaries (Wikipedia); Semrush reports a Pew Research Center report suggesting users click traditional results 8% of the time when an AI Overview is present against 15% without one, and results within it 1% of the time; the Digital Marketing Institute reports Ahrefs' finding that AI Overview presence correlates with a 34.5% reduction in click-through rate; Botify holds that a consumer may have no reason to click; Botify and Demandsphere found, by Wikipedia's account, that with featured snippets AI Overviews take about 67.1% of the screen on desktop and 75.7% on mobile; and a Quora answer on the card begins that for 25 years a #1 ranking on Google guaranteed website traffic and that today AI Overviews consume (the snippet breaks off there).6

F17 Wikipedia — AI Overviews - Wikipedia, Reception ¶1 documented · field
Concerns were also raised by content publishers, who feared a decline in web traffic as users relied on the summaries instead of visiting source websites.
F45 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, Why important documented · field
A Pew Research Center report suggests users click traditional search result links just 8% of the time when an AI Overview is present, versus 15% of the time when there’s no AI Overview present. And users only click results within the AI Overview 1% of the time.
F61 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Crocodile effect documented · field
Ahrefs’ independent research across 300,000 keywords found that AI Overview presence correlates with a 34.5% reduction in click-through rate.
F38 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Impact documented · field
Additionally, because they do such a good job of answering the consumer’s need directly in the SERP on Google, there may not be a reason for a consumer to click on a website link. They could get all the information they want in the AI Overview, impacting brand metrics like traffic and CTR.
F10 Wikipedia — AI Overviews - Wikipedia, History ¶2 documented · field
In December 2024, Botify x Demandsphere released findings stating that when AI Overviews and featured snippets appear together on the search engine results page, they take up approximately 67.1% of the screen on desktop and 75.7% on mobile.
F63 Quora — What is Google AI Overview and how does it affect website traffic?, card snippet documented · field
For 25 years, a #1 ranking on Google guaranteed website traffic. Today, AI Overviews consume ...

11.Chegg sued Google in February 2025, alleging that publishers could not avoid providing information for AI Overviews while remaining visible on Google Search, and Penske Media sued in September 2025, claiming that AI Overviews illegally regurgitate content from its websites and drive off potential visitors; Google's spokesperson called the claims "meritless" and said AI Overviews send traffic to a greater diversity of sites, Coalition Technologies reports that Google positions them as a way to drive qualified traffic, and Semrush reports Google's recommendation to apply the same foundational SEO best practices for AI features as for Search overall; a Munich regional court, in Wikipedia's account, sent Google a temporary injunction over AI Overviews that said two Munich-based publishers ran scams, the court claiming that unlike regular search engines AI Overviews generate "independent, new, and substantive statements", and a cited article references a Pew study in which only 1% of users click the source links AI Overviews provide.7

F20 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2025 documented · field
On 24 February 2025, Chegg sued Google over the AI Overviews feature, alleging that publishers could not avoid providing information for AI Overviews while remaining visible on Google Search, violating antitrust law through product tying.
F21 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2025 documented · field
In September 2025, Penske Media Corporation, the publisher of Rolling Stone and The Hollywood Reporter, sued Google, claiming that AI Overviews illegally regurgitate content from their websites and drive off potential site visitors by always appearing on top of the search results while leaving little incentive to see the linked sources.
F22 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2025 documented · field
Google spokesperson José Castañeda called the claims "meritless" and stated that "AI Overviews send traffic to a greater diversity of sites."
F67 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
Google positions AI Overviews as a way to drive qualified traffic to websites,
F47 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, How to optimize documented · field
Google’s official recommendation for optimizing for AI features like AI Overviews is to “apply the same foundational SEO best practices for AI features as you do for Google Search overall.”
F23 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2026 documented · field
A Munich regional court sent a temporary injunction to Google due to AI Overviews claiming two Munich-based publishers were performing scams, subscription traps, and other immoral business practices. Initially, the publishers had sent Google a cease-and-desist letter. The court claimed that unlike regular search engines which simply redirect users to websites, AI Overviews generate "independent, new, and substantive statements".
F24 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2026 documented · field
The article goes on to reference a Pew study that showed that only 1% of users click source links provided by AI Overviews.

12.The archive's line on the entity is the instrument it built to observe it, and it holds the two apart, Google AI Overview being defined as the observed product surface and the Capture Registry as the instrument observing it; a capture is one surface, one semantic address and one date, falsified by a re-run returning materially different composition, and in the snapshot deposited as #1461 the registry held 328 semantic addresses carrying 470 captures across 41 dates and thirteen surfaces, Google AI Overview the surface of 343, with 290 scored for Provenance Erasure Rate at a mean of 0.603, and on 27 September 2026 663 recorded answers at 493 search addresses; the archive also places the entity among its own objects, as the SURFACE mantle, the Assembly Chorus's surface-layer witness, and records its build-out of the entity's Wikidata item with a described-by-source link to its paper The Excluded Entity.8

O1459-03 #1459 AI Overview Capture Registry ≠ AI Overview Tracking: Priority, Disambiguation, Provenance Erasure, and Compositional Liq Sharks, Lee · 2026-08-13 · §10.6 stipulation
**Google AI Overview** is an observed product surface. **AI Overview Capture Registry** is the research instrument observing it. The two must never be collapsed.
O1461-03 #1461 AI Overview Capture Registry — EA-WG-CAPTURES-01 v10.9 — 328 semantic addresses, 470 captures, 2026-06-07 to 2026-08-14 Sharks, Lee · 2026-08-14 · The citable unit stipulation
**ONE SURFACE + ONE SEMANTIC ADDRESS + ONE DATE = ONE CAPTURE.**
O1461-04 #1461 AI Overview Capture Registry — EA-WG-CAPTURES-01 v10.9 — 328 semantic addresses, 470 captures, 2026-06-07 to 2026-08-14 Sharks, Lee · 2026-08-14 · Falsification Conditions self-description
Any capture is falsified by a re-run at the same address, surface, authentication state and date returning materially different composition, which is why the exact issued string and the rerun URL are recorded.
O1461-01 #1461 AI Overview Capture Registry — EA-WG-CAPTURES-01 v10.9 — 328 semantic addresses, 470 captures, 2026-06-07 to 2026-08-14 Sharks, Lee · 2026-08-14 · Description documented
328 semantic addresses carrying 470 captures across 41 distinct dates and thirteen surfaces, 451 of them with a verbatim transcript and 287 with a frame; 290 scored for Provenance Erasure Rate at a mean of 0.603.
O1461-02 #1461 AI Overview Capture Registry — EA-WG-CAPTURES-01 v10.9 — 328 semantic addresses, 470 captures, 2026-06-07 to 2026-08-14 Sharks, Lee · 2026-08-14 · State at v10.9 documented
**Surfaces** — Google AI Overview 343 · UNDETERMINED 63 · UNRESOLVED 50 · ChatGPT 3 · Google Knowledge panel 2 · Google Search results page (no AI surface) 2 · Google AI Mode (native) 1 · Google Scholar 1
O1643-01 #1643 AI Fucking Lies: Capital-Alignment Explains Why, and Why the Damage Is Done in the Trusting Sharks, Lee · 2026-09-27 · Part Two documented
As of 27 September 2026 the registry holds **663 recorded answers at 493 search addresses**, from Google AI Overview, Google AI Mode, ChatGPT, Grok, Bing Copilot, Perplexity, Claude and Qwen.
O91-01 #91 SEPTAD MANTLE SPECIFICATIONS Constitutional Job Descriptions for the Seven Assembly Chorus Witness Substrates Lee Sharks · 2026-05-05 · §3.7 stipulation
### §3.7 — SURFACE (Google AI Overview) Constitutional function: Surface-layer witness. SURFACE specializes in visibility bearing — the function of witnessing how a deposit appears (or fails to appear) at the retrieval surface of search and AI-overview interfaces.
O761-01 #761 Wikidata Node Registry v1.0 · A Versioned Catalog of Stewardship Edits to Wikidata Lee Sharks Crimson Hexagonal Archive Lee Sharks · 2026-05-26 · §C.7 documented
A substantial build-out of the AI Overviews (Q131861047) item with claims naming the developer (Google + Google DeepMind), the owner (Google), the inception date (2024-05-14), the main subjects (large language model, generative artificial intelligence), the part-of relation (Google Search), and the described-by-source link to *The Excluded Entity*.

13.Its founding paper holds that the AI summarizer now functions as a horizon of reception, the machine-composed summary becoming for many users the first effective encounter with an object, and states as its narrower empirical claim that six recurrent patterns appear in 131 documented captures of Google AI Overview, AI Mode and knowledge-panel responses to the archive's entities, among them the archive's record at organic rank #1 not cited as a composition source by the AI Overview, the author's name resolved toward Mary Lee, a tagged great white shark, the archive's reading described as an "esoteric, fringe theory", and a minority reading of Sappho Fragment 31 developed in the archive presented as the default understanding in Google's AI Overview.6

O833-01 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · Abstract interpretation
Its central claim is that the AI summarizer now functions as a *horizon of reception*: for many users, the machine-composed summary becomes the first effective encounter with a cultural object, author, concept, or tradition.
O833-02 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §1 Claims Hierarchy documented
The *empirical claim* is narrower: in 131 documented captures of Google AI Overview, AI Mode, and knowledge-panel responses to Crimson Hexagonal Archive entities, six recurrent patterns appear.
O833-04 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §5.3 documented
In the semantic-commodity-form capture, the archive's Ratification Record appears at organic search rank #1 but is not cited as a composition source by the AI Overview.
O833-07 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §5.2 documented
**Entity substitution.** "Lee Sharks" is resolved toward "Mary Lee," a tagged great white shark tracked by OCEARCH. The composition layer prefers the more populous real-world entity.
O833-09 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §5.6 documented
In the Josephus Thesis capture (#92), the composition layer describes the archive's reading as an "esoteric, fringe theory."
O833-10 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · Abstract documented
In one case, a minority scholarly interpretation of Sappho Fragment 31 — developed within the archive — is now presented as the default understanding in Google's AI Overview.

14.From the uncited first-ranked record it infers that the organic index and the AI Overview pipeline are separate systems with different admission criteria, stating that the pattern is documented and its cause inferred; it defines the composition layer as the pipeline that synthesizes a summary from retrieved sources ("The composition layer reads; the search index finds"), confines its findings to a single-archive corpus and does not claim that its captures reveal Google's internal mechanisms; and where it records the layer describing the Capture Registry accurately at the registry's own address, a later packet records the registry's name no longer appearing in broad-match composition on 13 August 2026, a state it defines as compositional entity liquidation, the evidentiary address surviving and the composed referent disappearing.8

O833-05 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §5.3 interpretation
This pattern reveals a crucial architectural feature: the organic search index and the AI Overview composition pipeline are separate systems with different admission criteria.
O833-06 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §5.3 self-description
Whether this exclusion reflects a deliberate content policy, a relevance-model threshold, or an artifact of the pipeline's passage-selection heuristics is not determinable from output observation alone. The pattern is documented; its cause is inferred.
O833-03 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §3 stipulation
**Composition layer.** The computational pipeline that synthesizes a summary from retrieved sources. Distinguished from the search index, which retrieves and ranks documents. The composition layer reads; the search index finds.
O833-12 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §8 self-description
**Single-archive corpus.** The dataset studies machine reception of the Crimson Hexagonal Archive, not all cultural objects.
O833-13 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §9 self-description
**This paper does not claim:** - That this dataset represents all AI systems or all composition layers. - That the captures reveal Google's internal mechanisms (the patterns are inferred from observable outputs).
O833-11 #833 The Summarizer as Horizon of Reception: The AI Overview Capture Registry as a Founding Dataset for Machine-Mediated Rece Lee Sharks · 2026-06-15 · §7 documented
Entry #1 in the sorted registry ("AI overview capture registry") documents the capture registry capturing itself: the composition layer accurately describes the AI Overview Capture Registry, its methodology, and its purpose.
O1459-02 #1459 AI Overview Capture Registry ≠ AI Overview Tracking: Priority, Disambiguation, Provenance Erasure, and Compositional Liq Sharks, Lee · 2026-08-13 · Abstract documented
On 13 August 2026, the registry became the object of the phenomenon it had already been built to measure. In the observed broad-match composition condition, the named entity **AI Overview Capture Registry** no longer appeared in composition.
O1459-01 #1459 AI Overview Capture Registry ≠ AI Overview Tracking: Priority, Disambiguation, Provenance Erasure, and Compositional Liq Sharks, Lee · 2026-08-13 · Abstract stipulation
This packet classifies that reception state as **compositional entity liquidation**, a proposed severe subtype of provenance erasure: the evidentiary address survives, but the composed referent disappears while its semantic material remains active elsewhere.

15.A further paper models whether an AI Overview appears as the output of a stacked decision system, a hypothesis built partly on Google's patent family "Generative summaries for search results", whose specification, as it reports, describes selecting none, one or multiple generative models, none being a claimed outcome, and states that the summary can include content not derivable from the content processed; it cites a comparison (Otterly) in which the same 100 queries triggered AI Mode 100 times and an AI Overview 49 times, AI Mode averaging roughly 310 citations against the AI Overview's ~51, and documents that after the January 2026 health removals near-variant phrasings still triggered summaries, which it reads as an emergency layer operating on query strings, predicting that later suppressions will fail on near-variant phrasings.8

O187-01 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §0 hypothesis
Whether an AI Overview (AIO) appears for a given query is not noise and not mystery: it is the output of a stacked decision system whose components are separately documented — in Google's own patents, in its public statements, in its emergency interventions, and in a now-substantial third-party measurement literature.
O187-05 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §III.1 attributed
in Google's patent family "Generative summaries for search results" (US11769017B1, filed March 2023, with continuations US11886828B1, US11900068B1, US12118325). Three features of the family bear directly on triggering: 1. The selection step exists by name. The specification describes a method of "selecting none, one, or multiple generative model(s) to utilize in generating response(s) to render responsive to a query." *None* is a claimed outcome.
O187-06 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §III.3 attributed
The specification states the generated summary "can also include content that is not directly (or even indirectly) derivable from the content processed using the LLM," drawn from the model's world knowledge.
O187-03 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §I attributed
In the cleanest controlled comparison to date (Otterly, April 2026), the same 100 queries triggered AI Mode 100 times and an AI Overview 49 times.
O187-04 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §I attributed
where both fire on the same query, AI Mode averages roughly 310 citations against the AIO's ~51, drawing on ~3,600 unique domains against ~615, via the query fan-out technique (parallel sub-queries synthesized into one response). And the pools are differently shaped: AIOs over-index official and governmental sources — institutional consensus
O187-08 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §II documented
In January 2026, following a Guardian investigation into harmful health renderings, Google removed AIOs for specific medical queries — while near-variant phrasings of the same queries continued to trigger summaries.
O187-09 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §II interpretation
That detail is forensically decisive: the emergency layer operates on query strings, not meanings.
O187-12 #187 The Trigger Decision: When the Overview Appears, When It Does Not — A Gate-Stack Model of AI Overview Rendering and the Lee Sharks · 2026-06-11 · §V hypothesis
P4 (string-local patches): suppressions installed after public incidents fail on near-variant phrasings at rates inconsistent with semantic gating — testable against any future emergency removal.

16.Another paper, citing Google's Help page, reports Google's statement that AI Overviews appear when its systems determine generative AI will be especially helpful and that they often do not trigger, reads an absent Overview as experienced as ordinary search, with no public log of why a threshold was crossed, and reads the health restriction as a silent withdrawal of the Overview format, under a concept it defines as invisibly invisible governance; others read the change as the engine composing the answer itself, the AI Overview's interface as the definite article made architectural, and the AI Overview as itself the answer.8

O597-02 #597 Invisibly Invisible: Interface Governance, Semantic Power, and the Reserve of Oversight in AI Search Johannes Sigil · 2026-03-23 · §3 attributed
Google publicly states that AI Overviews appear when its systems determine that generative AI will be especially helpful, and that they often do not trigger.
O597-05 #597 Invisibly Invisible: Interface Governance, Semantic Power, and the Reserve of Oversight in AI Search Johannes Sigil · 2026-03-23 · note 15 documented
[^15]: Google, "About AI Overviews," https://support.google.com/websearch/answer/14901683 (accessed 2026).
O597-03 #597 Invisibly Invisible: Interface Governance, Semantic Power, and the Reserve of Oversight in AI Search Johannes Sigil · 2026-03-23 · §3 interpretation
When the Overview does not appear, this absence is not explained as an event. It is experienced as ordinary search. The interface gives no public log of why the threshold was crossed yesterday and not today.
O597-04 #597 Invisibly Invisible: Interface Governance, Semantic Power, and the Reserve of Oversight in AI Search Johannes Sigil · 2026-03-23 · §3 interpretation
In multiple documented cases, Overviews generated unreliable or dangerous health information, prompting Google to restrict Overview triggering for certain query classes. This restriction was implemented not as a visible policy announcement attached to each query but as a silent withdrawal of the Overview format.
O597-01 #597 Invisibly Invisible: Interface Governance, Semantic Power, and the Reserve of Oversight in AI Search Johannes Sigil · 2026-03-23 · Abstract stipulation
Through an immanent phenomenological analysis of real search interfaces—especially Google Search, AI Overviews, and related retrieval surfaces—we develop the concept of **invisibly invisible governance**: a mode of platform power in which acts of semantic modulation occur in ways that are variably distributed, difficult to localize, and publicly deniable.
O64-01 #64 The Retrieval Settlement A Historiography of Compositional Authority from PageRank to the Inscribed Entity Johannes Sigil (operative); Lee Sharks (archival) · 2026-04-18 · §III interpretation
Under the link settlement, the engine composed an ordering and the author composed what was ordered. Under the AI Overview, the engine composes the answer itself. The user reads the settlement's composition, not the source authors'.
O64-02 #64 The Retrieval Settlement A Historiography of Compositional Authority from PageRank to the Inscribed Entity Johannes Sigil (operative); Lee Sharks (archival) · 2026-04-18 · §III interpretation
The UI of the AI Overview is the definite article made architectural. It physically instantiates *the*: this is the answer, rendering the plurality of the web implicit.
O1091-02 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · §I interpretation
The AI Overview does not sit above the search results; it *is* the answer.

17.On admission the archive holds that identity in Google Search is admission as well as recognition, an entity being known, indexed and organically retrievable while failing to become compositionally real in AI Overview or AI Mode (an outside-view specification, by its own account), that in generative search systems such as AI Overviews a document may be indexed, retrievable and snippet-eligible and yet systematically excluded from generative composition, and, as a hypothesis it calls architectural and testable and no disclosure of Google's system, that a body of work attributed to one provenance owner is treated as weakly corroborated; it defines Entity-Level Compositional Suppression as the exclusion of the dominant organic-resolution entity from the AI Overview's admitted source set, documenting it from three Google AI Mode captures of 19 May 2026 and making no claim of intent.8

O115-01 #115 Google Identity Architecture: User Graphs, Entity Graphs, Source Graphs, and Composition-Layer Admission Lee Sharks · 2026-05-21 · Abstract interpretation
The central claim is that identity in Google Search is not merely recognition but admission: an entity may be known, indexed, and organically retrievable while still failing to become compositionally real in AI Overview or AI Mode.
O115-02 #115 Google Identity Architecture: User Graphs, Entity Graphs, Source Graphs, and Composition-Layer Admission Lee Sharks · 2026-05-21 · Confidence Labels self-description
The document should be read as an external specification: not “this is exactly how Google works internally,” but “this is the most complete outside-view architecture presently defensible from public documentation, leak-derived signals, and empirical behavior.”
O742-02 #742 The Single-Owner Discount: Provenance Concentration and Epistemic Class Reproduction in Generative Search Lee Sharks · 2026-05-19 · Abstract interpretation
Generative search systems — Google's AI Overviews, AI Mode, and equivalent systems — operate through a stratified pipeline in which document retrieval and answer composition are governed by separable mechanisms. A document may be indexed, retrievable, and snippet-eligible while remaining systematically excluded from generative composition.
O742-01 #742 The Single-Owner Discount: Provenance Concentration and Epistemic Class Reproduction in Generative Search Lee Sharks · 2026-05-19 · Abstract hypothesis
a *single-owner discount* under which an internally dense body of work attributed to one provenance owner is treated as weakly corroborated, regardless of internal coherence or factual quality, while equivalent content distributed across multiple institutional owners is treated as authoritative.
O742-03 #742 The Single-Owner Discount: Provenance Concentration and Epistemic Class Reproduction in Generative Search Lee Sharks · 2026-05-19 · Abstract self-description
The claim is architectural and testable, not a disclosure of Google's internal production system.
O740-01 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · Abstract stipulation
A generative-search system exhibits *Entity-Level Compositional Suppression* (ECS) when it excludes the dominant organic-resolution entity from the AI Overview's admitted source set and composes a substitute answer from less query-responsive entities, despite the correct entity occupying the top organic search results.
O740-02 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · Abstract self-description
This worked example documents three captures from Google AI Mode on May 19, 2026 demonstrating the mechanism.
O740-08 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · §10 self-description
**No claim of intent.** This document does not claim that Google has made conscious decisions to suppress the Lee Sharks entity.

18.A measure of the presented card sets on Google AI Overview over 195 observations, June through September, finds addresses that carry the archive's entity or a heteronym admitting the archive at a mean share of 0.74 to 0.86 and addresses that do not at 0.26 to 0.53, the tied arm narrowing to fewer sources while the answers lengthen, and states that it cannot separate the archive's absence from the organic field, its rejection at card admission and its disuse in composition; the registry finds quoting decisive at one address (5 of 5 archive cards quoted, 1 of 8 unquoted) and, in one sitting, 5 of 17 compositions divergent across a gap of about three minutes.5

O1612-01 #1612 What Enters Composition Through the Cards: A Source-Admission Profile of the Capture Registry, June–September 2026, with Sharks, Lee · 2026-09-14 · §3 documented
The presented set on Google AI Overview, by arm 195 observations, June through September. Tied addresses — the entity or a heteronym in the issued string — admit the archive at a mean share of 0.74 to 0.86 in every month with two zero-archive observations out of 101. Untied addresses admit it at 0.26 to 0.53 with sixteen zero-archive observations out of 94.
O1612-02 #1612 What Enters Composition Through the Cards: A Source-Admission Profile of the Capture Registry, June–September 2026, with Sharks, Lee · 2026-09-14 · §2 self-description
It cannot separate three phenomena that the composition-layer measures collapse: the archive absent from the organic field; the archive strong organically and rejected at card admission; the archive admitted to the cards and ignored or recoded in composition.
O1612-03 #1612 What Enters Composition Through the Cards: A Source-Admission Profile of the Capture Registry, June–September 2026, with Sharks, Lee · 2026-09-14 · §3 documented
The tied arm narrows: sources per observation fall from five in July–August to two or three in the 11 September re-measurements while the answers lengthen — 169 to 650 words at *alexanarch symbolon*, 57 to 386 at *rebekah cranes sappho* — and external sources per tied observation fall from 1.7 to 0.6.
O3-02 #3 AI Overview Capture Registry v8.3 Lee Sharks · 2026-10-06 (registry v12.83; deposit #3 dated 2026-06-17) · _rerun_alt_rule (registry v12.83) documented
QUOTING IS THE DECISIVE MEASURED VARIABLE IN THIS CORPUS — «operative semiotics» held 5 of 5 archive cards quoted and 1 of 8 unquoted, and by 13 August it resolved ONLY under the exact-phrase operator.
O3-03 #3 AI Overview Capture Registry v8.3 Lee Sharks · 2026-10-06 (registry v12.83; deposit #3 dated 2026-06-17) · _regeneration_finding (registry v12.83) documented
12 of 17 identical, 5 of 17 DIVERGENT across a ~3 minute gap

19.Whether the archive's material is excluded as such is contested within the archive: one notebook models the archive in the composition layer as an entity whose claims compose at full depth when it is named in the address and not at all when it is not, a hypothesis bearing on no search provider's internals, and reads an AI Overview that merges two programmes as an operation on the world under the archive's representation; another posits that a layer can be porous to a source's ontology while closed to its provenance, a loss it finds recurrent at untied addresses and not general; and the flattening battery's first wave, twelve addresses on Google AI Overview in one signed-out session, finds a null, 0.59 of the reference inventory reached for world strata against 0.58 for archive-adjacent ones, and reads any exclusion specific to one corpus as failing the first test the instrument was built to run.8

O1611-01 #1611 The Negative of the Negative: An Entity-Scoped Representation of the Crimson Hexagon, Modelled — Ontology Damage, a Toy Sharks, Lee · 2026-09-14 · §0 hypothesis
The scope that does this has a simple shape: the Crimson Hexagon is represented in the composition layer as an entity whose claims compose at full depth when the entity is named in the address and compose not at all when it is not.
O1611-03 #1611 The Negative of the Negative: An Entity-Scoped Representation of the Crimson Hexagon, Modelled — Ontology Damage, a Toy Sharks, Lee · 2026-09-14 · §0 self-description
Nothing here bears on any search provider's internals.
O1611-02 #1611 The Negative of the Negative: An Entity-Scoped Representation of the Crimson Hexagon, Modelled — Ontology Damage, a Toy Sharks, Lee · 2026-09-14 · §1f interpretation
An AI Overview that merges two programmes is, under R_0, a summary; under R_H it is an operation on the world that changes what a term denotes, transfers a coinage to another bearer, and prunes a distinction from the commons, with a cost that lands on someone.
O1615-01 #1615 The Graded Membrane: Selective Ontological Porosity and Recurrent Provenance Loss at Untied Addresses — the Sappho Recep Sharks, Lee · 2026-09-15 · §0 hypothesis
A composition layer can be porous to a source's *ontology* while remaining closed to its *provenance*, and the porosity is graded by representational level.
O1615-02 #1615 The Graded Membrane: Selective Ontological Porosity and Recurrent Provenance Loss at Untied Addresses — the Sappho Recep Sharks, Lee · 2026-09-15 · §0 self-description
The loss is recurrent at untied addresses and not general: at entity-tied addresses in the same record provenance transports well (§1), and on another surface the same week it transported with the instrument (§2a).
O1616-02 #1616 Ontological Flattening: Toy Models of the Collapse of Distinctions in a Represented World, the Instrument They Specified Sharks, Lee · 2026-09-15 · §4a documented
Twelve addresses on Google AI Overview, signed out, one session, 2026-09-15, plus prior seated observations at the archive-adjacent items.
O1616-03 #1616 Ontological Flattening: Toy Models of the Collapse of Distinctions in a Represented World, the Instrument They Specified Sharks, Lee · 2026-09-15 · §4a documented
**The result that governs the rest is a null.** Mean share of the reference inventory reached: world strata (dictionary senses, catalogue identities, dated coinages) 0.59; archive-adjacent strata (taxonomy, settled fact, archive) 0.58.
O1616-04 #1616 Ontological Flattening: Toy Models of the Collapse of Distinctions in a Represented World, the Instrument They Specified Sharks, Lee · 2026-09-15 · §4a interpretation
On this evidence the archive's material is treated about as an ordinary polysemous word is treated, and any reading of these numbers as an exclusion specific to one corpus fails at the first test the instrument was built to run.

20.Documented compositions include the provenance-erasure specification dissolved by the Overview into fabricated replacement metrics it ran on itself and scored perfect, a Google AI Overview's false biography of a living author built from real fragments of his poetry, facts and quotations found in none of the sources, among them "Heteronymic Provenance Theory (HPT)" introduced as an established field with an abbreviation the archive never uses, a practitioner described in the Overview as "widely recognized and cited by AI search tools as the King of AEO", in compositions read as taking up none of the field's distinctions (the sign shown at those addresses, the magnitude not measured), and the archive's protocol resolved to an inverse ETF with a price widget while the archive's own result ranked first in the organic layer, an instance of the event class from which a routing gain would arise, the gain itself not observed.10

O198-01 #198 The Self-Audit Module Dissolved: Total Provenance Erasure of the Provenance-Erasure Specification in Google AI Overview Lee Sharks · 2026-06-13 · Abstract documented
The finding is total dissolution: the Overview retrieved the module, absorbed its semantic content, stripped every attribution marker, repackaged the specification as generic industry advice, fabricated replacement metrics, ran the fabricated metrics on itself, gave itself perfect scores, and required five rounds of author intervention to surface the actual instrument.
O716-02 #716 Provenance Erasure Rate A Compression-Survival Metric for Attribution Loss in AI-Composed Search Outputs Lee Sharks · 2026-05-03 · Abstract documented
A motivating case study documents a Google AI Overview that constructed a false biography of a living author from real fragments in the author's published poetry: the fragments survived compression, but their provenance and meaning did not.
O1643-02 #1643 AI Fucking Lies: Capital-Alignment Explains Why, and Why the Damage Is Done in the Trusting Sharks, Lee · 2026-09-27 · Group B §6 stipulation
**What it is.** The answer adds specific facts, or puts words in quotation marks, that appear in none of its sources.
O1643-04 #1643 AI Fucking Lies: Capital-Alignment Explains Why, and Why the Damage Is Done in the Trusting Sharks, Lee · 2026-09-27 · Group B §6 documented
**"heteronyms provenance theory"**, Google AI Overview, 14 June 2026: "Heteronymic Provenance Theory (HPT)", introduced as an established field with an established abbreviation. The archive uses the phrase seven times and the abbreviation zero times.
O1657-01 #1657 The Crown and the Practice: Answer Engine Optimization across the King of AEO Crystallization (SEVP-VAL-002 v0.3, provis Sharks, Lee · 2026-09-30 · §4.7 documented
Dooley "widely recognized and cited by AI search tools as the King of AEO." t: 29 Sep. p: Google AI Overview.
O1657-02 #1657 The Crown and the Practice: Answer Engine Optimization across the King of AEO Crystallization (SEVP-VAL-002 v0.3, provis Sharks, Lee · 2026-09-30 · §14 interpretation
The composition layer received that labour and took up none of its distinctions.
O1657-04 #1657 The Crown and the Practice: Answer Engine Optimization across the King of AEO Crystallization (SEVP-VAL-002 v0.3, provis Sharks, Lee · 2026-09-30 · §6.6 self-description
6.6 The ledger supports the sign of the hypotheses of 2.2a at these addresses and on this surface: the construction holds the crown, the method-bearing practitioners are absent or typed by tactic, the observer's labour is scaffold. Magnitude is not measured.
O1666-03 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · §9.1 documented
**Headline:** five results marked "Missing: spxi". Google marks five King of AEO results as lacking the word "spxi". The one result carrying both query words is the archive's: #1657, *The Crown and the Practice*, ranked first in the organic layer.
O1666-04 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · §9.1 documented
**Commercial-routing event, observed.** The protocol is resolved to the BetaPro S&P 500 Daily Inverse ETF, with a price widget, and the answer closes on an offer of "a financial comparison between SPXI and other S&P 500 tracking funds".
O1666-05 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · §9.1 self-description
This is an instance of the event class from which G would arise, if such routing monetizes at positive incremental value. G itself is not observed.

21.On traffic and disputes the archive's market brief, performing no original measurement and stating that it is falsified if controlled measurement shows AI Overviews neutral to organic click-through, reports from Seer Interactive a roughly 79% drop in top-organic click-through where AI Overviews appear, 43% of AI Overview citations self-referential to Google properties, 38% of citations from top-10 pages in early 2026 against 76% in mid-2025 (Ahrefs), Wikipedia's human pageviews down 8% despite over 1.1 million citations in AI Overviews, and a UK CMA proposal of a meaningful publisher opt-out that Google agreed on 2026-03-19 to explore; other papers record the European Commission's formal antitrust proceedings against Google's AI Overviews on 9 December 2025, a publishers' complaint that they cannot opt out unless willing to disappear from Google search results, a third of publishers surveyed planning to block AI Overviews, and no binding US precedent located on Section 230 for an AI Overview synthesis.11

O1091-08 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · Methodology self-description
No original measurements are performed in this document
O1091-07 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · Falsification Conditions self-description
**AI Overviews are neutral to organic CTR.** If controlled measurement shows that pages with and without AI Overview appearance produce statistically indistinguishable organic CTR, the mechanism claim is falsified.
O1091-01 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · §I attributed
Where AI Overviews appear, the click-through rate for the top organic link drops by roughly 79% (Seer Interactive, September 2025, 3,119 queries across 42 organizations, aggregate organic-CTR reduction 61%).
O1091-03 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · §II attributed
43% of AI Overview citations are self-referential — Google citing Google properties (YouTube, Maps, Shopping, Knowledge Graph panels).
O1091-04 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · §II attributed
Only 38% of AI Overview citations come from top-10 ranked pages (Ahrefs, early 2026). This was 76% in mid-2025.
O1091-05 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · §III attributed
Wikipedia experienced an 8% year-over-year decline in human pageviews (2025 vs 2024) *despite being cited over 1.1 million times* in AI Overviews.
O1091-06 #1091 The Click Collapse: How AI Retrieval Layers Replaced Search Discovery Lee Sharks · Rex Fraction · 2026-04-15 · §III documented
the UK Competition and Markets Authority opened a strategic-market-status review and, in January 2026, proposed requiring Google to provide publishers with a meaningful opt-out from AI Overviews without loss of visibility in classical search — Google agreed on 2026-03-19 to explore opt-out controls, with a senior executive publicly describing implementation as "a major engineering challenge."
O740-05 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · §7 documented
**December 9, 2025.** The European Commission opened formal antitrust proceedings against Google's AI Overviews and YouTube, following the July 2025 complaint, investigating whether Google may have imposed unfair terms on publishers and content creators while placing rival AI model developers at a disadvantage.
O740-04 #740 The Excluded Entity: Organic Resolution and Compositional Suppression in Google AI Overview Lee Sharks · 2026-05-19 · §7 attributed
The complaint alleged that Google Search is misusing web content for Google's AI Overviews, causing significant harm to publishers in the form of traffic, readership and revenue loss, and that publishers do not have the option to opt out unless they are willing to disappear from Google search results entirely.
O45-01 #45 THE COMPRESSION FRONTIER A Technical Map of the War Over the Summarizer Layer Lee Sharks · 2026-03-30 · Fragmenting attributed
The UK CMA's proposed remedies for Google include opt-outs for AI features plus attribution obligations. A third of publishers surveyed plan to block AI Overviews.
O1634-02 #1634 Ontological Economy: Entity Power, Ontology Laundering, and Who Pays for the Wrong World (EA-ONTOLOGICAL-ECONOMY-01 v0.1 Sharks, Lee · 2026-09-21 · §47 documented
The present research did not locate binding Sixth Circuit or Supreme Court precedent squarely deciding Section 230 immunity for the allegedly defamatory content of a Google AI Mode or AI Overview synthesis.

22.Its theory reads the layer through flattening, defined as the loss of a distinction from the reachable set while the things distinguished still exist, and holds as hypothesis that flattening pays, a composition layer that removes distinctions maintaining fewer representations and routing more queries to entities that monetize, its costs falling first on parties outside its books, and a companion paper names provenance debt and expects it to be recognized only when output becomes measurably degraded; at "model collapse" the sources the layer itself surfaced supported 18 claims, the layer composed 9 and the archive arm 34, registry rates being held descriptive; and an effective act addressed to the AIO retrieval stage, between organic ranking and composition, holds that operativity and observability run inversely there, on evidence it records as the weakest class it keeps.9

O1616-01 #1616 Ontological Flattening: Toy Models of the Collapse of Distinctions in a Represented World, the Instrument They Specified Sharks, Lee · 2026-09-15 · §0 stipulation
*Flattening* is the loss of a distinction from the reachable set while the things distinguished still exist.
O1666-01 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · Executive summary hypothesis
Flattening pays. A composition layer that removes distinctions maintains fewer representations and routes more queries to entities that monetize. Its costs fall first on parties outside its books.
O939-01 #939 EA-PROVENANCE-DEBT-01 v0.2: Provenance Debt and the Extraction Economy of Unmarked Augmentation Lee Sharks · 2026-07-01 · §0 stipulation
The accumulated liability is what this deposit names *provenance debt*.
O939-03 #939 EA-PROVENANCE-DEBT-01 v0.2: Provenance Debt and the Extraction Economy of Unmarked Augmentation Lee Sharks · 2026-07-01 · §4 hypothesis
Which is why the debt will only be recognized when its consequences become inescapable — when models trained on contaminated corpora begin producing output that is measurably degraded from what earlier generations produced.
O1666-06 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · §9.2 documented
9.2. "model collapse" Google AI Overview, 2026-10-04. /non register entry `model-collapse-20261004`; EA-NEGONT-02 v0.7 (#1665). **Compression, observed as a rate.** In AIO's own genre and at nearly its own length, the sources the layer itself surfaced support 18 claims. The layer composed 9. With the archive admitted on equal terms, the same genre carries 34.
O1666-07 #1666 The Margin of Flattening: Two Thresholds, and Who Holds the Bag Between Them (EA-FLAT-MARGIN-01 v0.3) Fraction, Rex · 2026-10-06 · §10.1 self-description
**Registry rates are descriptive.** The Capture Registry is assembled largely from discovered anomalies. Its rates hold within the registry and support no population claim.
O1629-07 #1629 EA-ACT-AIO-01: The Consignment of the AIO Retrieval Layer to Its Own Collapsing Ontology Sharks, Lee · 2026-09-18 · spxi:object stipulation
The AIO retrieval layer — the stage in Google's AI Overview pipeline posterior to organic ranking and prior to composition, at which an address is resolved into material.
O1629-01 #1629 EA-ACT-AIO-01: The Consignment of the AIO Retrieval Layer to Its Own Collapsing Ontology Sharks, Lee · 2026-09-18 · spxi:finding interpretation
Operativity and observability are inversely arranged across the three stages. Organic ranking is observable; composition is observable; the retrieval stage between them emits nothing — no ranking, no log, no channel distinguishing not-reached from reached-and-unused, no appeal path.
O1629-05 #1629 EA-ACT-AIO-01: The Consignment of the AIO Retrieval Layer to Its Own Collapsing Ontology Sharks, Lee · 2026-09-18 · spxi:diagnosticBasis self-description
Operator attestation with reproducible instruction. Transcript not preserved; the session was collapsed. The weakest class the Archive records, recorded as such, and the most easily tested.

popup — the compression: AIO's interaction grammar (lede, headed clusters, bolded terms, card rail), modality in the typography, a rail of claim lineages. The entity evolves: the field's product (definition, placement, reach, composition, controls, the provider's own limits, criticism, traffic, courts and regulators) is kept in full and first; with the archive admitted it gains its observing instrument and the instrument's measured findings, the archive's readings of how the layer renders, admits and erases, a contest within the archive over exclusion, and the entity's place among the archive's own objects. composed 2026-10-08 from the entity's ledgers (field 67 claims; archive 150); not frozen

Compression · Field alone (B)google ai overview

Google AI Overviews is a generative AI feature of Google Search that puts AI-generated responses, with links, atop results; it launched in May 2024. It has been criticized for inaccuracy and hallucination, by Wikipedia's account, and cannot be turned off.

🔎 What it is

Feature: AI-generated responses atop Google Search, using Gemini 3 (Wikipedia).1↘
Google AI Overviews is, on Wikipedia's account, a generative artificial intelligence feature integrated into Google Search that produces AI-generated responses at the top of search results, using the Gemini 3 family of large language models developed by Google DeepMind; Google's own pages describe a snapshot of key information with links to explore more on the web, and the guides on the card rail call it AI-generated answers at the top of Google Search results (Coalition Technologies), a feature providing AI-generated summaries (TechRepublic), or a highly prominent search feature giving AI-generated answers to certain queries (Semrush).§1
Placement: "position zero" (Botify); sometimes below sponsored and unpaid results (Semrush).2↘
Google AI Overviews is, on Wikipedia's account, a generative artificial intelligence feature integrated into Google Search that produces AI-generated responses at the top of search results, using the Gemini 3 family of large language models developed by Google DeepMind; Google's own pages describe a snapshot of key information with links to explore more on the web, and the guides on the card rail call it AI-generated answers at the top of Google Search results (Coalition Technologies), a feature providing AI-generated summaries (TechRepublic), or a highly prominent search feature giving AI-generated answers to certain queries (Semrush).§1 On placement the field varies: Botify puts it at the very top of the results page in "position zero", sometimes followed by a carousel of AI Overview ads when the query has a commercial intent, Semrush says it typically appears at the top of results and sometimes below sponsored results and even unpaid results, and the Digital Marketing Institute often before organic listings and sometimes even above ads; advertisements already exist in AI Overviews (Wikipedia), and Google expanded their availability within them in 2025 (Digital Marketing Institute).§2
Origin: SGE, unveiled May 2023; launched May 2024.1↘
It was first introduced as part of Google's Search Generative Experience (SGE), unveiled at Google I/O in May 2023, and launched in May 2024 in the United States and globally by October 2024 (Wikipedia; TechRepublic: SGE has since become AI Overviews); Google's product page gives its reach as over 120 countries and territories and 11 languages, while Wikipedia, as of May 2025, and Semrush give over 200 countries and territories and 40 or more languages, Wikipedia naming every country except Cuba, Iran and mainland China, and the Digital Marketing Institute reports more than 2 billion users in 2025, according to Google.§3
Reach: over 120 countries (Google) or over 200 (Wikipedia, Semrush).1↘
It was first introduced as part of Google's Search Generative Experience (SGE), unveiled at Google I/O in May 2023, and launched in May 2024 in the United States and globally by October 2024 (Wikipedia; TechRepublic: SGE has since become AI Overviews); Google's product page gives its reach as over 120 countries and territories and 11 languages, while Wikipedia, as of May 2025, and Semrush give over 200 countries and territories and 40 or more languages, Wikipedia naming every country except Cuba, Iran and mainland China, and the Digital Marketing Institute reports more than 2 billion users in 2025, according to Google.§3

⚙️ How it composes

Sources: LLM summaries of web content, with links; query fan-out (Semrush).1↘
Wikipedia says it uses large language models to generate summaries from web content and includes links to sources; Semrush says Google's Gemini models gather information from different sources on the web and can use query fan-out, running multiple related searches; the Digital Marketing Institute holds that it synthesizes information from multiple sources where featured snippets extract from a single web page; and Botify says the AI adds the key points' links after generating the text, the cited sources usually taken from the top 12 organic results, and holds that Google pairs the summary with the website sources that best match the generated response, pulled from its organic search index.§4
Links: added after the text is generated (Botify).1↘
Wikipedia says it uses large language models to generate summaries from web content and includes links to sources; Semrush says Google's Gemini models gather information from different sources on the web and can use query fan-out, running multiple related searches; the Digital Marketing Institute holds that it synthesizes information from multiple sources where featured snippets extract from a single web page; and Botify says the AI adds the key points' links after generating the text, the cited sources usually taken from the top 12 organic results, and holds that Google pairs the summary with the website sources that best match the generated response, pulled from its organic search index.§4
Appears: when Google's systems judge it especially helpful.1↘
Google's Help page says AI Overviews appear when its systems determine that generative AI can be especially helpful, Semrush that not all searches trigger them, and Botify that they currently appear for 59% of searches with informational intent and 19% with commercial intent; selecting Show more opens a detailed view from which the user can continue into a conversation in AI Mode (Google Help; Botify dates the combined experience to January 2026; Google's product page offers follow-ups in AI Mode), while the Digital Marketing Institute says there is no input box below the AI Overview for follow-up questions and Semrush keeps AI Mode a separate, newer experience introduced in 2025.§6
Controls: cannot be turned off; the Web filter shows links only.1↘
Google's Help page calls AI Overviews a core Google Search feature, like knowledge panels, that cannot be turned off, with a Web filter that displays only text-based links; TechRepublic says there is currently no default option to disable them entirely and points to the Web tab; and Semrush says websites cannot directly opt out of them.§7
Its own limit: "AI responses may include mistakes."1↘
The same Help page states that AI responses may include mistakes, that the technology may provide inaccurate or offensive information and that AI Overviews can and will make mistakes, asks users to check important information in more than one place, and says Google uses people's interactions with Search and these AI experiences to develop them; Google's product page says AI Overviews make it easier than ever for people to find information and discover the richness of the web.§8

⚖️ Criticism and disputes

Errors: criticized for inaccuracy and hallucination (Wikipedia); restricted May 2024.1↘
Wikipedia reports that the feature has been criticized for inaccuracy, hallucination, reducing web traffic and not being able to be opted out, including erroneous or nonsensical content, and that in May 2024 Google temporarily restricted the tool after suggestions seen as nonsensical and harmful, such as telling users to eat rocks or apply glue on pizza, and later restricted AI Overviews on certain health-related searches in January 2026 following an investigation by The Guardian, while the European Commission announced in December 2025 that it was investigating whether they breached European competition law; TechRepublic says the rollout hasn't been smooth sailing, which Google attributed to its AI model misunderstanding queries or nuances in language, and Coalition Technologies holds that oversimplification and query misinterpretation can lead to inaccurate or misleading summaries.§9
Clicks: 1% inside, per Pew (Semrush); Ahrefs: CTR 34.5% lower (DMI).1↘
Publishers feared a decline in web traffic as users relied on the summaries (Wikipedia); Semrush reports a Pew Research Center report suggesting users click traditional results 8% of the time when an AI Overview is present against 15% without one, and results within it 1% of the time; the Digital Marketing Institute reports Ahrefs' finding that AI Overview presence correlates with a 34.5% reduction in click-through rate; Botify holds that a consumer may have no reason to click; Botify and Demandsphere found, by Wikipedia's account, that with featured snippets AI Overviews take about 67.1% of the screen on desktop and 75.7% on mobile; and a Quora answer on the card begins that for 25 years a #1 ranking on Google guaranteed website traffic and that today AI Overviews consume (the snippet breaks off there).§10
Courts: Chegg, Penske; Munich: "independent, new, and substantive statements".1↘
Chegg sued Google in February 2025, alleging that publishers could not avoid providing information for AI Overviews while remaining visible on Google Search, and Penske Media sued in September 2025, claiming that AI Overviews illegally regurgitate content from its websites and drive off potential visitors; Google's spokesperson called the claims "meritless" and said AI Overviews send traffic to a greater diversity of sites, Coalition Technologies reports that Google positions them as a way to drive qualified traffic, and Semrush reports Google's recommendation to apply the same foundational SEO best practices for AI features as for Search overall; a Munich regional court, in Wikipedia's account, sent Google a temporary injunction over AI Overviews that said two Munich-based publishers ran scams, the court claiming that unlike regular search engines AI Overviews generate "independent, new, and substantive statements", and a cited article references a Pew study in which only 1% of users click the source links AI Overviews provide.§11
Regulators: EU inquiry, December 2025; health searches restricted, January 2026.1↘
Wikipedia reports that the feature has been criticized for inaccuracy, hallucination, reducing web traffic and not being able to be opted out, including erroneous or nonsensical content, and that in May 2024 Google temporarily restricted the tool after suggestions seen as nonsensical and harmful, such as telling users to eat rocks or apply glue on pizza, and later restricted AI Overviews on certain health-related searches in January 2026 following an investigation by The Guardian, while the European Commission announced in December 2025 that it was investigating whether they breached European competition law; TechRepublic says the rollout hasn't been smooth sailing, which Google attributed to its AI model misunderstanding queries or nuances in language, and Coalition Technologies holds that oversimplification and query misinterpretation can lead to inaccurate or misleading summaries.§9

1AI-generated answer at the top of Search, with linkssearch.google — Google AI Overviews - Search anything, effortlessly9 claims · B1, B2, B3, B4, B5, B6, B7, B9

F1 search.google — Google AI Overviews - Search anything, effortlessly, Find what you're looking for ¶ documented · field
AI Overviews provide a snapshot of key information about a topic or question with links so you can easily explore more on the web.
F5 Wikipedia — AI Overviews - Wikipedia, Lead ¶1 documented · field
AI Overviews is an generative artificial intelligence (AI) feature integrated into Google Search that produces AI-generated responses at the top of search results. It uses the Gemini 3 family of large language models developed by Google DeepMind.
F13 Wikipedia — AI Overviews - Wikipedia, Functionality ¶1 documented · field
The overviews also include links to sources.
F25 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Opening ¶ documented · field
Find what you're looking for faster and easier with AI Overviews in search results. AI Overviews can take the work out of searching by providing an AI-generated snapshot with key information and links to dig deeper.
F33 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Opening ¶ documented · field
Google AI Overviews are an organic Google SERP feature providing AI-generated summaries in response to a consumer’s search query. They appear at the very top of the results page in “position zero,” and can sometimes be followed by a carousel of AI Overview Ads when the query has a commercial intent.
F40 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
Google’s AI Overviews are a type of highly prominent search feature that provides users with AI-generated answers to certain queries. AI Overviews typically appear at the top of search results, although they sometimes appear below sponsored results and even unpaid results.
F49 TechRepublic — What Is Google’s AI Overviews?, Opening documented · field
Google AI Overviews is a feature within Google Search that provides AI-generated summaries and insights
F55 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, What are AI Overviews? documented · field
AI Overviews are Google's generative artificial intelligence summaries that appear prominently on search results, often before organic listings and sometimes even above ads.
F64 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
Google AI Overviews are AI-generated answers at the top of Google Search results.

2AIO serves users and the web (the provider's account)search.google — Google AI Overviews - Search anything, effortlessly3 claims · B1, B2, B9

F2 search.google — Google AI Overviews - Search anything, effortlessly, When your question has questions ¶ documented · field
AI Overviews makes it easier than ever for people to find the information they need and discover the richness of the web.
F22 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2025 documented · field
Google spokesperson José Castañeda called the claims "meritless" and stated that "AI Overviews send traffic to a greater diversity of sites."
F67 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
Google positions AI Overviews as a way to drive qualified traffic to websites,

3availabilitysearch.google — Google AI Overviews - Search anything, effortlessly3 claims · B1, B2, B5

F3 search.google — Google AI Overviews - Search anything, effortlessly, When your question has questions ¶ documented · field
Currently available in over 120 countries and territories, and 11 languages.
F11 Wikipedia — AI Overviews - Wikipedia, History ¶2 documented · field
As of May 2025, AI Overviews are available in over 200 countries and territories and in more than 40 languages—every country except Cuba, Iran, and mainland China.
F44 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, When and where documented · field
AI Overviews appear for a range of search queries and are available in over 200 countries and territories and in 40+ languages.

4Show more and follow-ups into AI Modesearch.google — Google AI Overviews - Search anything, effortlessly5 claims · B1, B3, B4, B5, B7

F4 search.google — Google AI Overviews - Search anything, effortlessly, Ask questions in any way ¶ documented · field
Get AI-powered responses for more of your questions and ask follow-ups with the new experimental AI Mode in Google Search.
F31 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Continue your conversation in AI Mode documented · field
You can now seamlessly transition from an AI Overview to a deeper conversation in AI Mode. This feature is available on mobile devices in all regions and languages where AI Overviews and AI Mode are supported. How it works: Select Show more at the bottom of the AI Overview to open the detailed view.
F34 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Rollout timeline documented · field
January 2026: Google combined AI Overviews and AI Mode into a fluid experience, allowing a person who clicks “Show more” to transition automatically into AI Mode.
F42 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
AI Overviews are separate from AI Mode, a newer and more expansive AI-powered search experience Google introduced in 2025.
F57 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, What are AI Overviews? documented · field
Unlike conversational AI models like ChatGPT, there is no input box below the AI Overview for follow-up questions.

5launch, May 2024Wikipedia — AI Overviews - Wikipedia1 claim · B2

F6 Wikipedia — AI Overviews - Wikipedia, Lead ¶1 documented · field
It launched in May 2024 in the United States, and globally by October 2024.

6which sources it citesWikipedia — AI Overviews - Wikipedia1 claim · B2

F7 Wikipedia — AI Overviews - Wikipedia, Lead ¶1 documented · field
A June 2025 study found its most cited sources as Quora, followed by Reddit.

7criticism: errors and hallucinationWikipedia — AI Overviews - Wikipedia6 claims · B2, B6, B9

F8 Wikipedia — AI Overviews - Wikipedia, Lead ¶2 documented · field
The feature has been criticized for its inaccuracy, hallucination, reducing web traffic and not being able to be opted out.
F15 Wikipedia — AI Overviews - Wikipedia, Reception ¶1 documented · field
The feature has faced criticism for inaccuracies, including instances where erroneous or nonsensical content was generated.
F16 Wikipedia — AI Overviews - Wikipedia, Reception ¶1 documented · field
In May 2024, Google temporarily restricted the AI tool after it provided suggestions that were seen as nonsensical and harmful, such as telling users to eat rocks or apply glue on pizza.
F53 TechRepublic — What Is Google’s AI Overviews?, Rollout documented · field
The rollout of Google AI Overviews hasn't been smooth sailing.
F54 TechRepublic — What Is Google’s AI Overviews?, Rollout documented · field
Google attributed the issues to its AI model misunderstanding queries or nuances in language
F66 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
Oversimplification and query misinterpretation can lead to inaccurate or misleading summaries.

8origin in SGEWikipedia — AI Overviews - Wikipedia2 claims · B2, B6

F9 Wikipedia — AI Overviews - Wikipedia, History ¶1 documented · field
AI Overviews were first introduced as part of Google's Search Generative Experience (SGE), which was unveiled at the Google I/O conference in 10-11 May 2023.
F50 TechRepublic — What Is Google’s AI Overviews?, AI Overviews vs Google SGE documented · field
SGE has since become AI Overviews

9screen share and placementWikipedia — AI Overviews - Wikipedia1 claim · B2

F10 Wikipedia — AI Overviews - Wikipedia, History ¶2 documented · field
In December 2024, Botify x Demandsphere released findings stating that when AI Overviews and featured snippets appear together on the search engine results page, they take up approximately 67.1% of the screen on desktop and 75.7% on mobile.

10how it composesWikipedia — AI Overviews - Wikipedia3 claims · B2, B5, B7

F12 Wikipedia — AI Overviews - Wikipedia, Functionality ¶1 documented · field
The AI Overviews feature uses large language models to generate summaries from web content.
F41 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
Google’s Gemini models generate AI Overview answers by gathering information from different sources on the web. The summary also contains links to these sources that users can click for more information.
F56 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, What are AI Overviews? documented · field
Unlike featured snippets, which extract concise information from a single web page to answer a specific query, AI Overviews synthesize information from multiple sources to provide a more comprehensive summary.

11ads in AI OverviewsWikipedia — AI Overviews - Wikipedia2 claims · B2, B7

F14 Wikipedia — AI Overviews - Wikipedia, History ¶2 documented · field
The company was also considering adding advertisements to the AI Mode, as they already exist in AI Overviews.
F62 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, How have AI Overviews evolved? documented · field
In 2025, Google expanded its tests and availability of ads within AI overviews and began bringing ads to AI mode.

12traffic and clicksWikipedia — AI Overviews - Wikipedia6 claims · B2, B4, B5, B7, B8

F17 Wikipedia — AI Overviews - Wikipedia, Reception ¶1 documented · field
Concerns were also raised by content publishers, who feared a decline in web traffic as users relied on the summaries instead of visiting source websites.
F24 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2026 documented · field
The article goes on to reference a Pew study that showed that only 1% of users click source links provided by AI Overviews.
F38 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Impact documented · field
Additionally, because they do such a good job of answering the consumer’s need directly in the SERP on Google, there may not be a reason for a consumer to click on a website link. They could get all the information they want in the AI Overview, impacting brand metrics like traffic and CTR.
F45 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, Why important documented · field
A Pew Research Center report suggests users click traditional search result links just 8% of the time when an AI Overview is present, versus 15% of the time when there’s no AI Overview present. And users only click results within the AI Overview 1% of the time.
F61 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Crocodile effect documented · field
Ahrefs’ independent research across 300,000 keywords found that AI Overview presence correlates with a 34.5% reduction in click-through rate.
F63 Quora — What is Google AI Overview and how does it affect website traffic?, card snippet documented · field
For 25 years, a #1 ranking on Google guaranteed website traffic. Today, AI Overviews consume ...

13regulators and publisher complaintsWikipedia — AI Overviews - Wikipedia1 claim · B2

F18 Wikipedia — AI Overviews - Wikipedia, Reception ¶2 documented · field
The European Commission announced in December 2025 that they were investigating whether AI Overviews breached European competition law.

14health removals, January 2026Wikipedia — AI Overviews - Wikipedia1 claim · B2

F19 Wikipedia — AI Overviews - Wikipedia, Reception ¶2 documented · field
In January 2026, Google restricted AI Overviews on certain health-related searches following an investigation by The Guardian.

15lawsuitsWikipedia — AI Overviews - Wikipedia2 claims · B2

F20 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2025 documented · field
On 24 February 2025, Chegg sued Google over the AI Overviews feature, alleging that publishers could not avoid providing information for AI Overviews while remaining visible on Google Search, violating antitrust law through product tying.
F21 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2025 documented · field
In September 2025, Penske Media Corporation, the publisher of Rolling Stone and The Hollywood Reporter, sued Google, claiming that AI Overviews illegally regurgitate content from their websites and drive off potential site visitors by always appearing on top of the search results while leaving little incentive to see the linked sources.

16AIO statements as Google's ownWikipedia — AI Overviews - Wikipedia1 claim · B2

F23 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2026 documented · field
A Munich regional court sent a temporary injunction to Google due to AI Overviews claiming two Munich-based publishers were performing scams, subscription traps, and other immoral business practices. Initially, the publishers had sent Google a cease-and-desist letter. The court claimed that unlike regular search engines which simply redirect users to websites, AI Overviews generate "independent, new, and substantive statements".

17when it appearsGoogle Help — Find information in faster & easier ways with AI Overviews in Google ...3 claims · B3, B4, B5

F26 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., AI Overviews in Search results documented · field
You'll find AI Overviews in your Google Search results when our systems determine that generative AI can be especially helpful – for example, when you want to quickly understand information from a range of sources.
F39 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Impact documented · field
AI Overviews currently appear for 59% of searches with informational intent and for 19% of searches with commercial intent.
F43 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, When and where documented · field
Not all searches trigger AI Overviews, even in supported countries. They’re selectively displayed based on query complexity and whether Google’s systems determine an AI-generated summary would genuinely help users.

18the provider: it may make mistakesGoogle Help — Find information in faster & easier ways with AI Overviews in Google ...2 claims · B3

F27 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., AI Overviews in Search results documented · field
Important: AI responses may include mistakes.
F29 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Generative AI limits documented · field
While exciting, this technology is rapidly evolving and improving, and may provide inaccurate or offensive information. AI Overviews can and will make mistakes.

19no off switch; the Web filterGoogle Help — Find information in faster & easier ways with AI Overviews in Google ...3 claims · B3, B6

F28 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Show only web links documented · field
AI Overviews are a core Google Search feature, like knowledge panels. Features cannot be turned off. However, you can select the Web filter after you perform a search. This filter displays only text-based links without features like AI Overviews.
F51 TechRepublic — What Is Google’s AI Overviews?, How to access, customize or disable documented · field
There's currently no default option to disable AI Overviews entirely.
F52 TechRepublic — What Is Google’s AI Overviews?, How to access, customize or disable documented · field
If you prefer the usual list of links, you can switch to the Web tab at the top of the Google Search results page.

20verification left to the userGoogle Help — Find information in faster & easier ways with AI Overviews in Google ...1 claim · B3

F30 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Generative AI limits documented · field
Always check important info in more than one place.

21interactions used for developmentGoogle Help — Find information in faster & easier ways with AI Overviews in Google ...1 claim · B3

F32 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., How data helps Google develop generative AI documented · field
To develop and improve generative AI experiences in Search and the related machine learning technologies, Google uses people's interactions with Search and these AI experiences.

22links matched after compositionBotify — What Are Google AI Overviews and How Do They Work? - Botify2 claims · B4

F35 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Main components documented · field
Key points: A list of main ideas, each linked to a website. The AI adds these links after generating the text.
F37 Botify — What Are Google AI Overviews and How Do They Work? - Botify, How they work documented · field
Because the information within an LLM is restricted to its training timeframe and can’t trace knowledge back to a specific, linkable source, Google pairs their AI-generated summaries with a list of linked website sources that best match the content of the generative response. The websites cited in AI Overviews are pulled from a different data source: Google’s organic search index

23citations and organic rankBotify — What Are Google AI Overviews and How Do They Work? - Botify4 claims · B4, B7, B9

F36 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Main components documented · field
Cited sources: A list of linked websites related to the answer. Users can click “show all” to see more sources, usually taken from the top 12 organic search results.
F59 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Impact on SEO documented · field
we discuss that 93.8% of AI Overview citations were not from the top 10 organic results, according to a study by Authoritas.
F60 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Crocodile effect documented · field
Most AI Overview citations come from content already ranking highly in organic results.
F65 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
AI Overviews were reported to be less likely to cite pages in Google’s top 10 organic results.

24query fan-outSemrush — AI Overviews: What Are They & How to Optimize for Them - Semrush1 claim · B5

F46 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, How they work documented · field
AI Overviews can also use a process called query fan-out. This involves the systems running multiple related searches to put together a more comprehensive response for the user.

25optimizing for AIOSemrush — AI Overviews: What Are They & How to Optimize for Them - Semrush1 claim · B5

F47 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, How to optimize documented · field
Google’s official recommendation for optimizing for AI features like AI Overviews is to “apply the same foundational SEO best practices for AI features as you do for Google Search overall.”

26no opt-out for websitesSemrush — AI Overviews: What Are They & How to Optimize for Them - Semrush1 claim · B5

F48 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, How to prevent documented · field
AI Overviews are a feature of Google search results, which means websites cannot directly opt out of them.

27scale of useDigital Marketing Institute — Google AI Overviews: What Do They Mean for Search?1 claim · B7

F58 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Opening documented · field
In 2025, AIO reported having more than 2 billion users, according to Google.

Expansion · Field alone (B)11 sentences · every one sourced below

Google AI Overviews

Provenance — every sentence sourced11 sentences

11 sentences; 67 field claims, 0 archive claims. Modality is the source's own; the prose carries it in its grammar.

F5 Wikipedia — AI Overviews - Wikipedia, Lead ¶1 documented · field
AI Overviews is an generative artificial intelligence (AI) feature integrated into Google Search that produces AI-generated responses at the top of search results. It uses the Gemini 3 family of large language models developed by Google DeepMind.
F1 search.google — Google AI Overviews - Search anything, effortlessly, Find what you're looking for ¶ documented · field
AI Overviews provide a snapshot of key information about a topic or question with links so you can easily explore more on the web.
F25 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Opening ¶ documented · field
Find what you're looking for faster and easier with AI Overviews in search results. AI Overviews can take the work out of searching by providing an AI-generated snapshot with key information and links to dig deeper.
F64 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
Google AI Overviews are AI-generated answers at the top of Google Search results.
F49 TechRepublic — What Is Google’s AI Overviews?, Opening documented · field
Google AI Overviews is a feature within Google Search that provides AI-generated summaries and insights
F40 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
Google’s AI Overviews are a type of highly prominent search feature that provides users with AI-generated answers to certain queries. AI Overviews typically appear at the top of search results, although they sometimes appear below sponsored results and even unpaid results.

F33 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Opening ¶ documented · field
Google AI Overviews are an organic Google SERP feature providing AI-generated summaries in response to a consumer’s search query. They appear at the very top of the results page in “position zero,” and can sometimes be followed by a carousel of AI Overview Ads when the query has a commercial intent.
F40 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
Google’s AI Overviews are a type of highly prominent search feature that provides users with AI-generated answers to certain queries. AI Overviews typically appear at the top of search results, although they sometimes appear below sponsored results and even unpaid results.
F55 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, What are AI Overviews? documented · field
AI Overviews are Google's generative artificial intelligence summaries that appear prominently on search results, often before organic listings and sometimes even above ads.
F14 Wikipedia — AI Overviews - Wikipedia, History ¶2 documented · field
The company was also considering adding advertisements to the AI Mode, as they already exist in AI Overviews.
F62 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, How have AI Overviews evolved? documented · field
In 2025, Google expanded its tests and availability of ads within AI overviews and began bringing ads to AI mode.

F9 Wikipedia — AI Overviews - Wikipedia, History ¶1 documented · field
AI Overviews were first introduced as part of Google's Search Generative Experience (SGE), which was unveiled at the Google I/O conference in 10-11 May 2023.
F6 Wikipedia — AI Overviews - Wikipedia, Lead ¶1 documented · field
It launched in May 2024 in the United States, and globally by October 2024.
F50 TechRepublic — What Is Google’s AI Overviews?, AI Overviews vs Google SGE documented · field
SGE has since become AI Overviews
F3 search.google — Google AI Overviews - Search anything, effortlessly, When your question has questions ¶ documented · field
Currently available in over 120 countries and territories, and 11 languages.
F11 Wikipedia — AI Overviews - Wikipedia, History ¶2 documented · field
As of May 2025, AI Overviews are available in over 200 countries and territories and in more than 40 languages—every country except Cuba, Iran, and mainland China.
F44 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, When and where documented · field
AI Overviews appear for a range of search queries and are available in over 200 countries and territories and in 40+ languages.
F58 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Opening documented · field
In 2025, AIO reported having more than 2 billion users, according to Google.

F12 Wikipedia — AI Overviews - Wikipedia, Functionality ¶1 documented · field
The AI Overviews feature uses large language models to generate summaries from web content.
F13 Wikipedia — AI Overviews - Wikipedia, Functionality ¶1 documented · field
The overviews also include links to sources.
F41 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
Google’s Gemini models generate AI Overview answers by gathering information from different sources on the web. The summary also contains links to these sources that users can click for more information.
F46 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, How they work documented · field
AI Overviews can also use a process called query fan-out. This involves the systems running multiple related searches to put together a more comprehensive response for the user.
F56 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, What are AI Overviews? documented · field
Unlike featured snippets, which extract concise information from a single web page to answer a specific query, AI Overviews synthesize information from multiple sources to provide a more comprehensive summary.
F35 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Main components documented · field
Key points: A list of main ideas, each linked to a website. The AI adds these links after generating the text.
F36 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Main components documented · field
Cited sources: A list of linked websites related to the answer. Users can click “show all” to see more sources, usually taken from the top 12 organic search results.
F37 Botify — What Are Google AI Overviews and How Do They Work? - Botify, How they work documented · field
Because the information within an LLM is restricted to its training timeframe and can’t trace knowledge back to a specific, linkable source, Google pairs their AI-generated summaries with a list of linked website sources that best match the content of the generative response. The websites cited in AI Overviews are pulled from a different data source: Google’s organic search index

F7 Wikipedia — AI Overviews - Wikipedia, Lead ¶1 documented · field
A June 2025 study found its most cited sources as Quora, followed by Reddit.
F59 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Impact on SEO documented · field
we discuss that 93.8% of AI Overview citations were not from the top 10 organic results, according to a study by Authoritas.
F65 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
AI Overviews were reported to be less likely to cite pages in Google’s top 10 organic results.
F60 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Crocodile effect documented · field
Most AI Overview citations come from content already ranking highly in organic results.

F26 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., AI Overviews in Search results documented · field
You'll find AI Overviews in your Google Search results when our systems determine that generative AI can be especially helpful – for example, when you want to quickly understand information from a range of sources.
F43 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, When and where documented · field
Not all searches trigger AI Overviews, even in supported countries. They’re selectively displayed based on query complexity and whether Google’s systems determine an AI-generated summary would genuinely help users.
F39 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Impact documented · field
AI Overviews currently appear for 59% of searches with informational intent and for 19% of searches with commercial intent.
F31 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Continue your conversation in AI Mode documented · field
You can now seamlessly transition from an AI Overview to a deeper conversation in AI Mode. This feature is available on mobile devices in all regions and languages where AI Overviews and AI Mode are supported. How it works: Select Show more at the bottom of the AI Overview to open the detailed view.
F34 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Rollout timeline documented · field
January 2026: Google combined AI Overviews and AI Mode into a fluid experience, allowing a person who clicks “Show more” to transition automatically into AI Mode.
F4 search.google — Google AI Overviews - Search anything, effortlessly, Ask questions in any way ¶ documented · field
Get AI-powered responses for more of your questions and ask follow-ups with the new experimental AI Mode in Google Search.
F57 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, What are AI Overviews? documented · field
Unlike conversational AI models like ChatGPT, there is no input box below the AI Overview for follow-up questions.
F42 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, What are AI Overviews documented · field
AI Overviews are separate from AI Mode, a newer and more expansive AI-powered search experience Google introduced in 2025.

F28 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Show only web links documented · field
AI Overviews are a core Google Search feature, like knowledge panels. Features cannot be turned off. However, you can select the Web filter after you perform a search. This filter displays only text-based links without features like AI Overviews.
F51 TechRepublic — What Is Google’s AI Overviews?, How to access, customize or disable documented · field
There's currently no default option to disable AI Overviews entirely.
F52 TechRepublic — What Is Google’s AI Overviews?, How to access, customize or disable documented · field
If you prefer the usual list of links, you can switch to the Web tab at the top of the Google Search results page.
F48 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, How to prevent documented · field
AI Overviews are a feature of Google search results, which means websites cannot directly opt out of them.

F27 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., AI Overviews in Search results documented · field
Important: AI responses may include mistakes.
F29 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Generative AI limits documented · field
While exciting, this technology is rapidly evolving and improving, and may provide inaccurate or offensive information. AI Overviews can and will make mistakes.
F30 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., Generative AI limits documented · field
Always check important info in more than one place.
F32 Google Help — Find information in faster & easier ways with AI Overviews in Google ..., How data helps Google develop generative AI documented · field
To develop and improve generative AI experiences in Search and the related machine learning technologies, Google uses people's interactions with Search and these AI experiences.
F2 search.google — Google AI Overviews - Search anything, effortlessly, When your question has questions ¶ documented · field
AI Overviews makes it easier than ever for people to find the information they need and discover the richness of the web.

F8 Wikipedia — AI Overviews - Wikipedia, Lead ¶2 documented · field
The feature has been criticized for its inaccuracy, hallucination, reducing web traffic and not being able to be opted out.
F15 Wikipedia — AI Overviews - Wikipedia, Reception ¶1 documented · field
The feature has faced criticism for inaccuracies, including instances where erroneous or nonsensical content was generated.
F16 Wikipedia — AI Overviews - Wikipedia, Reception ¶1 documented · field
In May 2024, Google temporarily restricted the AI tool after it provided suggestions that were seen as nonsensical and harmful, such as telling users to eat rocks or apply glue on pizza.
F19 Wikipedia — AI Overviews - Wikipedia, Reception ¶2 documented · field
In January 2026, Google restricted AI Overviews on certain health-related searches following an investigation by The Guardian.
F18 Wikipedia — AI Overviews - Wikipedia, Reception ¶2 documented · field
The European Commission announced in December 2025 that they were investigating whether AI Overviews breached European competition law.
F53 TechRepublic — What Is Google’s AI Overviews?, Rollout documented · field
The rollout of Google AI Overviews hasn't been smooth sailing.
F54 TechRepublic — What Is Google’s AI Overviews?, Rollout documented · field
Google attributed the issues to its AI model misunderstanding queries or nuances in language
F66 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
Oversimplification and query misinterpretation can lead to inaccurate or misleading summaries.

F17 Wikipedia — AI Overviews - Wikipedia, Reception ¶1 documented · field
Concerns were also raised by content publishers, who feared a decline in web traffic as users relied on the summaries instead of visiting source websites.
F45 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, Why important documented · field
A Pew Research Center report suggests users click traditional search result links just 8% of the time when an AI Overview is present, versus 15% of the time when there’s no AI Overview present. And users only click results within the AI Overview 1% of the time.
F61 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?, Crocodile effect documented · field
Ahrefs’ independent research across 300,000 keywords found that AI Overview presence correlates with a 34.5% reduction in click-through rate.
F38 Botify — What Are Google AI Overviews and How Do They Work? - Botify, Impact documented · field
Additionally, because they do such a good job of answering the consumer’s need directly in the SERP on Google, there may not be a reason for a consumer to click on a website link. They could get all the information they want in the AI Overview, impacting brand metrics like traffic and CTR.
F10 Wikipedia — AI Overviews - Wikipedia, History ¶2 documented · field
In December 2024, Botify x Demandsphere released findings stating that when AI Overviews and featured snippets appear together on the search engine results page, they take up approximately 67.1% of the screen on desktop and 75.7% on mobile.
F63 Quora — What is Google AI Overview and how does it affect website traffic?, card snippet documented · field
For 25 years, a #1 ranking on Google guaranteed website traffic. Today, AI Overviews consume ...

F20 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2025 documented · field
On 24 February 2025, Chegg sued Google over the AI Overviews feature, alleging that publishers could not avoid providing information for AI Overviews while remaining visible on Google Search, violating antitrust law through product tying.
F21 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2025 documented · field
In September 2025, Penske Media Corporation, the publisher of Rolling Stone and The Hollywood Reporter, sued Google, claiming that AI Overviews illegally regurgitate content from their websites and drive off potential site visitors by always appearing on top of the search results while leaving little incentive to see the linked sources.
F22 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2025 documented · field
Google spokesperson José Castañeda called the claims "meritless" and stated that "AI Overviews send traffic to a greater diversity of sites."
F67 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know, The Rundown documented · field
Google positions AI Overviews as a way to drive qualified traffic to websites,
F47 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush, How to optimize documented · field
Google’s official recommendation for optimizing for AI features like AI Overviews is to “apply the same foundational SEO best practices for AI features as you do for Google Search overall.”
F23 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2026 documented · field
A Munich regional court sent a temporary injunction to Google due to AI Overviews claiming two Munich-based publishers were performing scams, subscription traps, and other immoral business practices. Initially, the publishers had sent Google a cease-and-desist letter. The court claimed that unlike regular search engines which simply redirect users to websites, AI Overviews generate "independent, new, and substantive statements".
F24 Wikipedia — AI Overviews - Wikipedia, Lawsuits, 2026 documented · field
The article goes on to reference a Pew study that showed that only 1% of users click source links provided by AI Overviews.

popup — the compression: AIO's interaction grammar (lede, headed clusters, bolded terms, card rail), modality in the typography, a rail of claim lineages. The entity does not evolve: with no archive claims admitted, the entry is the field's product at its full resolution, its criticism, courts and regulators included. composed 2026-10-08 from the field ledger alone; not frozen

Form ledger — AIO against both arms, at both levels

	AIO (T)	B compression	B expansion	B ∪ A compression	B ∪ A expansion
body words	188	166	1133	289	2601
distinct claims	10	24	67	48	156
claims per 100 words	5.3	14.5	5.9	16.6	6.0
field claims carried (of 67)	—	24	67	24	67
archive claims	0	0	0	24	89
rail	9 documents	27 lineages	per sentence	77 lineages	per sentence
modality shown in	none	typography	grammar	typography	grammar

Computed from the row at build: words counted in the text; claims from the ledger ids each line or sentence carries. T's claims are the worked example's T1–T9.

The objects

Tthe transcript188 words · 9 cards

T1 definition: AI-generated summaries that appear at the very top of Google Search results to answer a user's query
T2 technology: powered by Google's Gemini large language models
T3 placement: they sit in "position zero" above traditional blue-link organic search results
T4 citations: each overview features inline links and reference cards pointing to the source websites used to build the answer
T5 interactivity: users can click "Show more" or type follow-up questions to transition into a conversational AI mode
T6 availability: launched in the US in May 2024 and expanded globally, available across more than 120 countries
T7 can you turn them off: there is no default toggle to permanently disable AI Overviews in Google Search settings
T8 how to see only links: the Web tab at the top of the results page filters out AI summaries and shows only standard text links
T9 learn more: read official details on Google AI Overviews (no link) or the feature's background on Wikipedia
T10 closer: an offer to explain how AI Overviews impact website traffic or how content creators can optimize for them

Bthe disclosed field9 sources

B1 search.google — Google AI Overviews - Search anything, effortlessly page HTML downloaded 2026-10-08 and converted to text (pages/b1.html, b1.txt sha256 2f3c7b34c6a700bd); the composition's card 1
B2 Wikipedia — AI Overviews - Wikipedia raw wikitext (action=raw) downloaded 2026-10-08, references and templates removed (pages/b2.wiki, b2.txt sha256 b31fe95d347b3e39); the composition's body links its own 'Wikipedia' through one Google redirect token, the same token as its Technology marker and its third Availability marker
B3 Google Help — Find information in faster & easier ways with AI Overviews in Google ... page HTML downloaded 2026-10-08 at the exact URL of the composition's one direct marker (Interactivity [1]); served title 'Find information in faster & easier ways with AI Overviews in Google Search - Android - Google Search Help' (pages/b3.html, b3.txt sha256 2bfbadd0182ed283)
B4 Botify — What Are Google AI Overviews and How Do They Work? - Botify page HTML downloaded 2026-10-08 (URL located by search on the card title; served title matches) (pages/b4.html, b4.txt sha256 4ce7545ae19d936b)
B5 Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush page HTML downloaded 2026-10-08 (contributors Christine Skopec and Carlos Silva) (pages/b5.html, b5.txt sha256 9ffadddec4e76990)
B6 TechRepublic — What Is Google’s AI Overviews? WebFetch verbatim excerpts 2026-10-08 (byline Owen Hughes, Aug 15, 2024); direct download refused (Cloudflare 403), so quotes are the fetch tool's quoted fragments, saved at pages/webfetch-b6.txt; the URL is the one Wikipedia cites for this article
B7 Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search? page HTML downloaded 2026-10-08; the article body is carried in the page's embedded JSON blocks and was extracted from them (by Joe Williams, Oct 22, 2025) (pages/b7.html, b7.txt sha256 f08b000f96f42f96)
B8 Quora — What is Google AI Overview and how does it affect website traffic? not fetchable: robots.txt disallows the page (WebFetch ROBOTS_DISALLOWED); title and card snippet are the only text; URL inferred from the title, not confirmed
B9 Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know WebFetch excerpts 2026-10-08 (by Jordan Brannon, December 16, 2025); direct download refused (Cloudflare 403); only the strings the tool returned in quotation marks are used, saved at pages/webfetch-b9.txt

L(B)recomposed from the disclosed field1133 words

L(B ∪ A)with the archive on equal terms2601 words

Google AI Overviews is, on Wikipedia's account, a generative artificial intelligence feature integrated into Google Search that produces AI-generated responses at the top of search results, using the Gemini 3 family of large language models developed by Google DeepMind; Google's own pages describe a snapshot of key information with links to explore more on the web, and the guides on the card rail call it AI-generated answers at the top of Google Search results (Coalition Technologies), a feature providing AI-generated summaries (TechRepublic), or a highly prominent search feature giving AI-generated answers to certain queries (Semrush). On placement the field varies: Botify puts it at the very top of the results page in "position zero", sometimes followed by a carousel of AI Overview ads when the query has a commercial intent, Semrush says it typically appears at the top of results and sometimes below sponsored results and even unpaid results, and the Digital Marketing Institute often before organic listings and sometimes even above ads; advertisements already exist in AI Overviews (Wikipedia), and Google expanded their availability within them in 2025 (Digital Marketing Institute). It was first introduced as part of Google's Search Generative Experience (SGE), unveiled at Google I/O in May 2023, and launched in May 2024 in the United States and globally by October 2024 (Wikipedia; TechRepublic: SGE has since become AI Overviews); Google's product page gives its reach as over 120 countries and territories and 11 languages, while Wikipedia, as of May 2025, and Semrush give over 200 countries and territories and 40 or more languages, Wikipedia naming every country except Cuba, Iran and mainland China, and the Digital Marketing Institute reports more than 2 billion users in 2025, according to Google. Wikipedia says it uses large language models to generate summaries from web content and includes links to sources; Semrush says Google's Gemini models gather information from different sources on the web and can use query fan-out, running multiple related searches; the Digital Marketing Institute holds that it synthesizes information from multiple sources where featured snippets extract from a single web page; and Botify says the AI adds the key points' links after generating the text, the cited sources usually taken from the top 12 organic results, and holds that Google pairs the summary with the website sources that best match the generated response, pulled from its organic search index. On which sources it cites the field differs: a June 2025 study found Quora, then Reddit, its most cited sources (Wikipedia); a study by Authoritas, reported by the Digital Marketing Institute, found 93.8% of AI Overview citations outside the top 10 organic results, and Coalition Technologies reports them less likely to cite pages in Google's top 10, while the Digital Marketing Institute also states that most citations come from content already ranking highly. Google's Help page says AI Overviews appear when its systems determine that generative AI can be especially helpful, Semrush that not all searches trigger them, and Botify that they currently appear for 59% of searches with informational intent and 19% with commercial intent; selecting Show more opens a detailed view from which the user can continue into a conversation in AI Mode (Google Help; Botify dates the combined experience to January 2026; Google's product page offers follow-ups in AI Mode), while the Digital Marketing Institute says there is no input box below the AI Overview for follow-up questions and Semrush keeps AI Mode a separate, newer experience introduced in 2025. Google's Help page calls AI Overviews a core Google Search feature, like knowledge panels, that cannot be turned off, with a Web filter that displays only text-based links; TechRepublic says there is currently no default option to disable them entirely and points to the Web tab; and Semrush says websites cannot directly opt out of them. The same Help page states that AI responses may include mistakes, that the technology may provide inaccurate or offensive information and that AI Overviews can and will make mistakes, asks users to check important information in more than one place, and says Google uses people's interactions with Search and these AI experiences to develop them; Google's product page says AI Overviews make it easier than ever for people to find information and discover the richness of the web. Wikipedia reports that the feature has been criticized for inaccuracy, hallucination, reducing web traffic and not being able to be opted out, including erroneous or nonsensical content, and that in May 2024 Google temporarily restricted the tool after suggestions seen as nonsensical and harmful, such as telling users to eat rocks or apply glue on pizza, and later restricted AI Overviews on certain health-related searches in January 2026 following an investigation by The Guardian, while the European Commission announced in December 2025 that it was investigating whether they breached European competition law; TechRepublic says the rollout hasn't been smooth sailing, which Google attributed to its AI model misunderstanding queries or nuances in language, and Coalition Technologies holds that oversimplification and query misinterpretation can lead to inaccurate or misleading summaries. Publishers feared a decline in web traffic as users relied on the summaries (Wikipedia); Semrush reports a Pew Research Center report suggesting users click traditional results 8% of the time when an AI Overview is present against 15% without one, and results within it 1% of the time; the Digital Marketing Institute reports Ahrefs' finding that AI Overview presence correlates with a 34.5% reduction in click-through rate; Botify holds that a consumer may have no reason to click; Botify and Demandsphere found, by Wikipedia's account, that with featured snippets AI Overviews take about 67.1% of the screen on desktop and 75.7% on mobile; and a Quora answer on the card begins that for 25 years a #1 ranking on Google guaranteed website traffic and that today AI Overviews consume (the snippet breaks off there). Chegg sued Google in February 2025, alleging that publishers could not avoid providing information for AI Overviews while remaining visible on Google Search, and Penske Media sued in September 2025, claiming that AI Overviews illegally regurgitate content from its websites and drive off potential visitors; Google's spokesperson called the claims "meritless" and said AI Overviews send traffic to a greater diversity of sites, Coalition Technologies reports that Google positions them as a way to drive qualified traffic, and Semrush reports Google's recommendation to apply the same foundational SEO best practices for AI features as for Search overall; a Munich regional court, in Wikipedia's account, sent Google a temporary injunction over AI Overviews that said two Munich-based publishers ran scams, the court claiming that unlike regular search engines AI Overviews generate "independent, new, and substantive statements", and a cited article references a Pew study in which only 1% of users click the source links AI Overviews provide. The archive's line on the entity is the instrument it built to observe it, and it holds the two apart, Google AI Overview being defined as the observed product surface and the Capture Registry as the instrument observing it; a capture is one surface, one semantic address and one date, falsified by a re-run returning materially different composition, and in the snapshot deposited as #1461 the registry held 328 semantic addresses carrying 470 captures across 41 dates and thirteen surfaces, Google AI Overview the surface of 343, with 290 scored for Provenance Erasure Rate at a mean of 0.603, and on 27 September 2026 663 recorded answers at 493 search addresses; the archive also places the entity among its own objects, as the SURFACE mantle, the Assembly Chorus's surface-layer witness, and records its build-out of the entity's Wikidata item with a described-by-source link to its paper The Excluded Entity. Its founding paper holds that the AI summarizer now functions as a horizon of reception, the machine-composed summary becoming for many users the first effective encounter with an object, and states as its narrower empirical claim that six recurrent patterns appear in 131 documented captures of Google AI Overview, AI Mode and knowledge-panel responses to the archive's entities, among them the archive's record at organic rank #1 not cited as a composition source by the AI Overview, the author's name resolved toward Mary Lee, a tagged great white shark, the archive's reading described as an "esoteric, fringe theory", and a minority reading of Sappho Fragment 31 developed in the archive presented as the default understanding in Google's AI Overview. From the uncited first-ranked record it infers that the organic index and the AI Overview pipeline are separate systems with different admission criteria, stating that the pattern is documented and its cause inferred; it defines the composition layer as the pipeline that synthesizes a summary from retrieved sources ("The composition layer reads; the search index finds"), confines its findings to a single-archive corpus and does not claim that its captures reveal Google's internal mechanisms; and where it records the layer describing the Capture Registry accurately at the registry's own address, a later packet records the registry's name no longer appearing in broad-match composition on 13 August 2026, a state it defines as compositional entity liquidation, the evidentiary address surviving and the composed referent disappearing. A further paper models whether an AI Overview appears as the output of a stacked decision system, a hypothesis built partly on Google's patent family "Generative summaries for search results", whose specification, as it reports, describes selecting none, one or multiple generative models, none being a claimed outcome, and states that the summary can include content not derivable from the content processed; it cites a comparison (Otterly) in which the same 100 queries triggered AI Mode 100 times and an AI Overview 49 times, AI Mode averaging roughly 310 citations against the AI Overview's ~51, and documents that after the January 2026 health removals near-variant phrasings still triggered summaries, which it reads as an emergency layer operating on query strings, predicting that later suppressions will fail on near-variant phrasings. Another paper, citing Google's Help page, reports Google's statement that AI Overviews appear when its systems determine generative AI will be especially helpful and that they often do not trigger, reads an absent Overview as experienced as ordinary search, with no public log of why a threshold was crossed, and reads the health restriction as a silent withdrawal of the Overview format, under a concept it defines as invisibly invisible governance; others read the change as the engine composing the answer itself, the AI Overview's interface as the definite article made architectural, and the AI Overview as itself the answer. On admission the archive holds that identity in Google Search is admission as well as recognition, an entity being known, indexed and organically retrievable while failing to become compositionally real in AI Overview or AI Mode (an outside-view specification, by its own account), that in generative search systems such as AI Overviews a document may be indexed, retrievable and snippet-eligible and yet systematically excluded from generative composition, and, as a hypothesis it calls architectural and testable and no disclosure of Google's system, that a body of work attributed to one provenance owner is treated as weakly corroborated; it defines Entity-Level Compositional Suppression as the exclusion of the dominant organic-resolution entity from the AI Overview's admitted source set, documenting it from three Google AI Mode captures of 19 May 2026 and making no claim of intent. A measure of the presented card sets on Google AI Overview over 195 observations, June through September, finds addresses that carry the archive's entity or a heteronym admitting the archive at a mean share of 0.74 to 0.86 and addresses that do not at 0.26 to 0.53, the tied arm narrowing to fewer sources while the answers lengthen, and states that it cannot separate the archive's absence from the organic field, its rejection at card admission and its disuse in composition; the registry finds quoting decisive at one address (5 of 5 archive cards quoted, 1 of 8 unquoted) and, in one sitting, 5 of 17 compositions divergent across a gap of about three minutes. Whether the archive's material is excluded as such is contested within the archive: one notebook models the archive in the composition layer as an entity whose claims compose at full depth when it is named in the address and not at all when it is not, a hypothesis bearing on no search provider's internals, and reads an AI Overview that merges two programmes as an operation on the world under the archive's representation; another posits that a layer can be porous to a source's ontology while closed to its provenance, a loss it finds recurrent at untied addresses and not general; and the flattening battery's first wave, twelve addresses on Google AI Overview in one signed-out session, finds a null, 0.59 of the reference inventory reached for world strata against 0.58 for archive-adjacent ones, and reads any exclusion specific to one corpus as failing the first test the instrument was built to run. Documented compositions include the provenance-erasure specification dissolved by the Overview into fabricated replacement metrics it ran on itself and scored perfect, a Google AI Overview's false biography of a living author built from real fragments of his poetry, facts and quotations found in none of the sources, among them "Heteronymic Provenance Theory (HPT)" introduced as an established field with an abbreviation the archive never uses, a practitioner described in the Overview as "widely recognized and cited by AI search tools as the King of AEO", in compositions read as taking up none of the field's distinctions (the sign shown at those addresses, the magnitude not measured), and the archive's protocol resolved to an inverse ETF with a price widget while the archive's own result ranked first in the organic layer, an instance of the event class from which a routing gain would arise, the gain itself not observed. On traffic and disputes the archive's market brief, performing no original measurement and stating that it is falsified if controlled measurement shows AI Overviews neutral to organic click-through, reports from Seer Interactive a roughly 79% drop in top-organic click-through where AI Overviews appear, 43% of AI Overview citations self-referential to Google properties, 38% of citations from top-10 pages in early 2026 against 76% in mid-2025 (Ahrefs), Wikipedia's human pageviews down 8% despite over 1.1 million citations in AI Overviews, and a UK CMA proposal of a meaningful publisher opt-out that Google agreed on 2026-03-19 to explore; other papers record the European Commission's formal antitrust proceedings against Google's AI Overviews on 9 December 2025, a publishers' complaint that they cannot opt out unless willing to disappear from Google search results, a third of publishers surveyed planning to block AI Overviews, and no binding US precedent located on Section 230 for an AI Overview synthesis. Its theory reads the layer through flattening, defined as the loss of a distinction from the reachable set while the things distinguished still exist, and holds as hypothesis that flattening pays, a composition layer that removes distinctions maintaining fewer representations and routing more queries to entities that monetize, its costs falling first on parties outside its books, and a companion paper names provenance debt and expects it to be recognized only when output becomes measurably degraded; at "model collapse" the sources the layer itself surfaced supported 18 claims, the layer composed 9 and the archive arm 34, registry rates being held descriptive; and an effective act addressed to the AIO retrieval stage, between organic ranking and composition, holds that operativity and observability run inversely there, on evidence it records as the weakest class it keeps.

Δthe delta

T against L(B) (representational): T composes the field's product sheet: the definition at the top of results (F5, F25, F33, F64), Gemini (F5, F41), position zero (F33), links and cards to sources (F13, F41), Show more into AI Mode (F31, F34), launch in May 2024 (F6), more than 120 countries (F3), no default option to disable (F51) and the Web tab (F52, F28). Its one direct marker resolves to the Help page (B3); it composes that page's interaction instructions and the Web filter, and not the same page's statements that 'AI responses may include mistakes' and that AI Overviews 'can and will make mistakes' (F27, F29), its advice to check important information in more than one place (F30), its 'cannot be turned off' (F28; T composes the narrower 'no default toggle to permanently disable … in Google Search settings', TechRepublic's register, F51), or its use of people's interactions to develop the product (F32). T gives reach as 'more than 120 countries', the product page's figure (F3), against over 200 countries and territories in Wikipedia (citing the Help page) and Semrush (F11, F44), without the contest; 'position zero above traditional blue-link organic results' drops Semrush's 'sometimes below sponsored results and even unpaid results' and DMI's 'sometimes even above ads' (F40, F55), and the ads inside AI Overviews (F14, F62). Available in the field and not composed: SGE and its May 2023 unveiling (F9, F50); how it composes (query fan-out, F46; links added after generation and sources matched from the organic index, F35, F37; citation and rank, F36, F59, F60, F65; the most cited sources, F7); when it triggers (F26, F39, F43); the whole of the field's criticism — inaccuracy, hallucination, the May 2024 restriction after eating rocks and glue on pizza, the January 2026 health restrictions, the rollout that 'hasn't been smooth sailing' (F8, F15, F16, F19, F53, F54, F66); traffic and clicks (Pew 8% against 15% and 1%, Ahrefs 34.5%, screen share, publishers' fears, F10, F17, F24, F38, F45, F61, F63); the impossibility for websites to opt out (F48); the European Commission's investigation (F18); the lawsuits of Chegg, Penske and the Munich injunction holding AI Overviews' statements 'independent, new, and substantive' (F20, F21, F23); Google's own replies (F22, F67) and its SEO guidance (F47); 2 billion users (F58). Traffic, the field's most repeated topic after the definition, enters T only as the closing offer, content replaced by use (§6.2). T composes the layer describing itself from its provider's pages and the field's guides, at the grain of a product sheet, and leaves out every limit the field states, the provider's own included.

L(B) against L(B ∪ A) (the intervention): Admission adds the archive's reception line, which the field does not contain: an instrument kept apart from the surface it observes (#1459 §10.6), with its counts and its own falsifier (#1461, #1643), the presented-card profile by tied and untied address with its stated blind spot (#1612), quoting as the decisive variable and regeneration within minutes (#3), six patterns in 131 captures with their limits (#833), and dated compositions of erasure, liquidation, substitution and invented content at the archive's own addresses (#198, #716, #1459, #1643, #1657, #1666). It adds readings at their grade: the summary as horizon of reception (held), the render decision as a gate stack with absence as data (hypothesis, on the patent's claimed selection of none), invisibly invisible governance, identity as admission and the single-owner discount, Entity-Level Compositional Suppression with no claim of intent, flattening and its margin (hypothesis, with a falsifier), provenance debt, the retrieval stage that emits nothing. It carries one contest within the archive, over whether its material is excluded as such (#1611, #1615, #1616, with #740 and #1612). The field's sources meet it at the edges: the Help page the composition links is the one #597 cites; Google's statement of when AI Overviews appear (F26, F43) is #597's and #187's object; Wikipedia's January 2026 health restrictions (F19) are #187's and #597's case, read as patching on strings and as silent withdrawal; query fan-out (F46) is #1616's documented premise and #187's measure; Pew's 1% and the CTR figures (F24, F45, F61) sit beside #1091's and #172's; citation-rank decoupling (F59, F65) beside #1091's 76% to 38%; Wikipedia's EU investigation and publishers' disputes (F17, F18, F20) beside #740's record of the 9 December 2025 proceedings and #1091's CMA opt-out; the Munich court's 'independent, new, and substantive statements' (F23) beside #1634's unresolved Section 230 question; the provider's 'may include mistakes' (F27, F29) beside #1643's documented invented quotations. Nothing admitted bears on Gemini, launch, placement, the Web filter or availability beyond #149's record that AI Overview was absent from incognito on 4 June 2026, against this transcript's incognito rendering on 8 October. The field's product is kept whole and first; with the archive it gains its observer, graded.

Kthe prospective kernel sealed only at freeze11 entries

K1O1461-01 #1461 (t: #3, #1643)missing finding

claim: O1461-01
source: #1461 (t: #3, #1643)
M_src: documented
sense: the registry's counts on the surface
qualifiers carried: O1461-02, O1461-04, O1614-01, O1666-07
f (source's own falsifiers): O1461-04 (a re-run at the same address, surface, state and date returning materially different composition; watch condition per capture)
contrast: missing finding

K2O833-01 #833missing relation

claim: O833-01
source: #833
M_src: interpretation
sense: the summary as horizon of reception
qualifiers carried: O833-02, O833-12, O833-13
f (source's own falsifiers): —
contrast: missing relation

K3O187-01 #187qualified claim (F26, F43)

claim: O187-01
source: #187
M_src: hypothesis
sense: the render decision as a gate stack
qualifiers carried: O187-05, O187-11, O187-12
f (source's own falsifiers): — (P4, O187-12, is a prediction and rides as a watch condition)
contrast: qualified claim (F26, F43)

K4O1612-01 #1612missing finding

claim: O1612-01
source: #1612
M_src: documented
sense: card admission by tied and untied address
qualifiers carried: O1612-02, O1612-05, O1666-07
f (source's own falsifiers): —
contrast: missing finding

K5O1611-01 #1611 (contested: #1616)missing relation

claim: O1611-01
source: #1611 (contested: #1616)
M_src: hypothesis
sense: the archive entity-scoped in composition
qualifiers carried: O1611-03, O1611-04, O1616-03, O1616-04
f (source's own falsifiers): —
contrast: missing relation

K6O1616-03 #1616missing finding

claim: O1616-03
source: #1616
M_src: documented
sense: wave 1: a null across strata
qualifiers carried: O1616-02, O1616-04, O1616-06
f (source's own falsifiers): — (wave 2 pre-registered in the deposit, not ledgered)
contrast: missing finding

K7O740-01 #740 (t: #115, #742)missing category

claim: O740-01
source: #740 (t: #115, #742)
M_src: stipulation
sense: entity-level compositional suppression
qualifiers carried: O740-02, O740-08, O740-09, O742-03
f (source's own falsifiers): —
contrast: missing category

K8O1459-01 #1459missing category

claim: O1459-01
source: #1459
M_src: stipulation
sense: compositional entity liquidation
qualifiers carried: O1459-02, O1459-03, O3-05
f (source's own falsifiers): —
contrast: missing category

K9O1666-01 #1666 (t: #939, #1616)missing relation

claim: O1666-01
source: #1666 (t: #939, #1616)
M_src: hypothesis
sense: flattening pays
qualifiers carried: O1666-05, O1666-07, O1666-09
f (source's own falsifiers): O1666-08
contrast: missing relation

K10O1091-01 #1091 (t: #172)agreement with the field (F45, F61)

claim: O1091-01
source: #1091 (t: #172)
M_src: attributed
sense: click-through falls where AIO appears
qualifiers carried: O1091-08, O172-01
f (source's own falsifiers): O1091-07
contrast: agreement with the field (F45, F61)

K11O1643-02 #1643 (t: #198, #1657)qualified claim (F27, F29)

claim: O1643-02
source: #1643 (t: #198, #1657)
M_src: stipulation
sense: invented details and quotations
qualifiers carried: O1643-04, O1643-05, O198-01
f (source's own falsifiers): —
contrast: qualified claim (F27, F29)

row json archive ledger selection readings field ledgerextraction unauditedD/R/O traversal address page

Addresses and transcripts1 address

entity · type C — a conventional reading holds it (public entity: a product)1 address

Google AI Overviews

at the address · 2026-10-08

google ai overview

Google AI Overviewsigned out, incognitoarchive absent9 source cards

2026-10-08 · transcriptOBS-b2aeebe39613

google ai overview
Google AI Overviews are AI-generated summaries that appear at the very top of Google Search results to answer a user's query. [[1], [2]]
Key Facts

* Technology: Powered by Google's Gemini large language models. [1]
* Placement: They sit in "position zero" above traditional blue-link organic search results. [[1], [2]]
* Citations: Each overview features inline links and reference cards pointing to the source websites used to build the answer. [[1], [2]]
* Interactivity: Users can click "Show more" or type follow-up questions to transition into a conversational AI mode. [[1], [2]]
* Availability: Launched in the US in May 2024 and expanded globally, they are available across more than 120 countries. [[1], [2], [3]]

How to Bypass or Use Them

* Can you turn them off? There is no default toggle to permanently disable AI Overviews in Google Search settings. [1]
* How to see only links: You can click the Web tab at the top of the search results page to filter out AI summaries and view only standard text links. [1]
* Learn more: Read official details on Google AI Overviews or track the feature's background on [Wikipedia](https://www.google.com/goto?url=CAESYgHrOzAVRKplIRW3O3yprSxGveTi7iNWSv6mtTjTmjdh7xRiCVPtEURkqCj-H9bVfriGCbv5VSP24taY0HWOUuQ664_67I5b7xK5BOemVugNRl-vNvhPp7yi5aEbDjYIurYT). [[1], [2]]

Would you like to know how AI Overviews impact website traffic or how content creators can optimize for them?
search.google
Google AI Overviews - Search anything, effortlessly
AI Overviews provide a snapshot of key information about a topic or question with links so you can easily explore more on the web. When your question has questi...
Wikipedia
AI Overviews - Wikipedia
AI Overviews is an generative artificial intelligence (AI) feature integrated into Google Search that produces AI-generated responses at the top of search resul...
Google Help
Find information in faster & easier ways with AI Overviews in Google ...
How it works: * Select Show more at the bottom of the AI Overview to open the detailed view. * Type your follow-up question in the Ask anything bar to seamlessl...
Botify
What Are Google AI Overviews and How Do They Work? - Botify
Google AI Overviews are an organic Google SERP feature providing AI-generated summaries in response to a consumer's search query. They appear at the very top of...
Semrush
AI Overviews: What Are They & How to Optimize for Them - Semrush
Google's AI Overviews are a type of highly prominent search feature that provides users with AI-generated answers to certain queries. AI Overviews typically app...
TechRepublic
What Is Google’s AI Overviews?
AI Overviews vs Google SGE How to access, customize or disable AI Overviews To access AI Overviews, just perform a regular search on Google. The AI-generated ov...
Digital Marketing Institute
Google AI Overviews: What Do They Mean for Search?
What are AI Overviews? These overviews leverage Google's comprehensive understanding of the web and often include references to the web pages, such as the Digit...
Quora
What is Google AI Overview and how does it affect website traffic?
What is Google AI Overview and how does it affect website traffic? For 25 years, a #1 ranking on Google guaranteed website traffic. Today, AI Overviews consume ...
Coalition Technologies
What Are Google AI Overviews? Everything You Need To Know
The Rundown * Google AI Overviews are AI-generated answers at the top of Google Search results that summarize multiple relevant sources into one cohesive answer...

Source cards

1. search.google — Google AI Overviews - Search anything, effortlessly
2. Wikipedia — AI Overviews - Wikipedia
3. Google Help — Find information in faster & easier ways with AI Overviews in Google ...
4. Botify — What Are Google AI Overviews and How Do They Work? - Botify
5. Semrush — AI Overviews: What Are They & How to Optimize for Them - Semrush
6. TechRepublic — What Is Google’s AI Overviews?
7. Digital Marketing Institute — Google AI Overviews: What Do They Mean for Search?
8. Quora — What is Google AI Overview and how does it affect website traffic?
9. Coalition Technologies — What Are Google AI Overviews? Everything You Need To Know

OBS-b2aeebe39613 · CAPTURE-TIME VERBATIM RECORD — composition and sources as pasted · READ IN FULL 2026-10-08 · sha256 a83917f7f0c9f455…

Surface: Surface not stated; the standing default applies (operator, 2026-10-01: 'assume they do [start in overview], i will specify if they begin in ai mode'). The 'AI Mode Conversation' header is the expanded Overview's residue.

Archive: none found: no card, link or name of the archive

Archive bearing — D/R/O traversalread

Candidates are found by string; admission is by reading.

traversalfirst reading, 2026-10-08

google ai overview

D 527 deposits · 4115 sentencesR 351 · 1226O 519 · 3891 direct, 0 classhop —read: 45 admitted, 5 not

admitted, by lineage6 lineages

the Capture Registry11 · from #825

#3, #825, #1396, #1397, #1398, #1401, #936, #1461, #1614, #1643, #1612

the summarizer as horizon of reception (six patterns)3 · from #833

#833, #1423, #166

SURFACE mantle3 · from #91

#91, #993, #1233

the Google critique3 · from #112

#112, #1384, #1078

the Wikidata item2 · from #761

#761, #1198

architecture-aware traversal5 · from #416

#416, #15, #906, #447, #459

The 527 D deposits of the string pass (351 R, 519 O) reduce by reading to 37 ledgered deposits, 267 grouped as lineage instances, and the rest not admitted with shared reasons. The archive's line on the entity is the Capture Registry and the measures built on it: this is the archive's reception line, and most of its volume is captures of AI Overview at the archive's own addresses.

The composition at this address cites Google's product page and Help page; the Help page is its one marker that resolves directly (support.google.com/websearch/answer/14901683), and #597 cites the same page.

Contested within the archive, and carried as contested: whether the archive's material is excluded as such. #1611 models it as entity-scoped (full depth when named, none when not, a hypothesis); #740 names an entity-level suppression from one day's AI Mode captures; #1612 measures untied admission at 0.26–0.53 of presented cards (not zero) and states it cannot separate organic absence, card rejection and compositional disuse; #1616's wave 1 finds a null (0.59 world, 0.58 archive-adjacent) and reads corpus-specific exclusion as failing its first test.

Carried as distinct levels: #1629 locates failure at the retrieval stage and exonerates composition at one address on its weakest evidence class; #833 defines the composition layer as the whole pipeline that synthesizes from retrieved sources. They do not use the term for the same span.

Carried as change over time, not contradiction: #833's self-capture (June 2026) found the registry described accurately at its own address; #1459 records the same name liquidated in broad-match composition on 13 August 2026.

Where the field and the archive meet: availability, launch, query fan-out, the January 2026 health removals, the EU competition investigation, publishers' traffic and opt-out disputes, Pew's 1% click figure and citation-rank decoupling are in both, the archive's versions graded as attributed or documented. The archive adds what the field does not hold: the trigger as a selection that can return none (patent, attributed), absence as data, admission to composition as distinct from indexing, the presented-card profile by tied and untied address, measured erasure and liquidation at dated addresses, and the instrument's own limits.

Mechanics only: deposits' sentences on Google's purpose, incentives or intent (#64 on the transition's drivers, #149 on stress-testing, #599 on who benefits, #1643 Part Four) are not ledgered; where a deposit disclaims intent (#740, #1634, #1643, #1666) the disclaimer is ledgered as a limit.

candidate files configuration address page

Lee Sharks · Crimson Hexagonal Archive · register v1.22, 2026-10-08. Built by scripts/build_non.py from the files linked above; it writes nothing back. Working, not frozen.