{ "row": "oecd-ai-incidents-monitor", "entity": "oecd-ai-incidents-monitor", "address": "oecd ai incidents and hazards monitor", "addresses": [ { "address": "oecd ai incidents and hazards monitor", "slug": "oecd-ai-incidents-and-hazards-monitor-aio-20261029", "addr_id": "ADDR-de46fae4a143", "obs_id": "OBS-23f50c1b7e04", "axn": "AXN:0AB2.CAPTURE.🗡️✏️🎹🔻🍃📖" } ], "epoch": "2026-10-29", "type": "C (public entity on which the archive bears: a monitor; an entity of the MMRS meta-aggregator)", "surface": "Google AI Overview (expanded)", "auth": "signed out, incognito", "status": { "procedure": "EA-NEGONT-02 v0.9, #1690: equal case, ρ = extends, read from the MMRS record and the capture record; archive selected by D/R/O (one deposit of two candidates admitted, #637 at O-class; #170 a homonym on the acronym; hop of 11 read and rejected; the class 'AI incident reporting and monitoring' matched nothing), with the MMRS record (31 claims) and the Capture Registry's findings and analysis (8 claims) ledgered as the archive's records of the entity; the field B by three origins (6 card sources for 6 cards, 1 own page beyond them, 2 Wikipedia records); the composer read the ledgers, never the transcript, except where the brief asks for T's claims and the delta", "ledger": "unaudited (§3.8: one extractor per ledger); field quotes 64/64 verified by script against saved text, PDF text, oEmbed title or card snippet; archive quotes 40/40 verified verbatim against the registry, census.json, articles.py and AXN-01ED; carried as marked: F32, F33 internally_inconsistent", "frozen": false, "breaches": [] }, "spec": "EA-NEGONT-02 v0.9, #1690", "objects": { "T": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/T1-oecd-ai-incidents-and-hazards-monitor-aio-20261029.txt", "sha256": "9c284bc639b5beb5f2e124db6cf0af4ee8b3095e22d51423073f32ab5cf3408f", "words": 219, "cards": 6, "words_note": "one transcript (219 words: the body from the answer's first line through the closer, inline links reduced to their text and citation markers removed, the card rail excluded; 6 cards). The path is the registry's transcript string, saved verbatim (its sha256 is the string's).", "claims": [ "T1-1 definition: 'a real-time tracking tool developed by the OECD.AI Policy Observatory in collaboration with the Patrick J. McGovern Foundation to monitor actual and potential risks posed by artificial intelligence' (marker: an opaque token, the one shared with the Real-Time Tracking line and the incident definition)", "T1-2 Real-Time Tracking: 'It continuously aggregates and analyzes world news using the Event Registry platform, processing over 150,000 news articles daily to detect AI-related events.'", "T1-3 Evidence Base: 'It provides policymakers, developers, and the public with empirical data on how AI risks materialize globally.'", "T1-4 Thematic Classification: 'Media reports are categorized into 14 distinct risk themes—including synthetic media, cyberattacks and fraud, autonomous vehicles, large language model (LLM) hallucinations, labour market disruptions, and child safety.'", "T1-5 AI Incident: 'An event where the development, use, or malfunction of an AI system results in actual harm (such as injury, health damage, critical infrastructure disruption, or human/fundamental rights violations).'", "T1-6 AI Hazard: 'An event or circumstance where the development or use of an AI system could plausibly lead to an AI incident (a near-harm or precursor event).'", "T1-7 access: 'You can explore live data, interactive filters, and recent tracking reports directly on the OECD AI Incidents Monitor.'", "T1-8 closer: 'Would you like to explore specific risk categories (such as labor market impacts or synthetic media) or look into the common reporting framework guidelines?'" ], "register": { "register": "data/EA-WG-CAPTURES-01.json", "slug": "oecd-ai-incidents-and-hazards-monitor-aio-20261029", "obs_id": "OBS-23f50c1b7e04" }, "transcripts": [ { "slug": "oecd-ai-incidents-and-hazards-monitor-aio-20261029", "address": "oecd ai incidents and hazards monitor", "addr_id": "ADDR-de46fae4a143", "obs_id": "OBS-23f50c1b7e04", "axn": "AXN:0AB2.CAPTURE.🗡️✏️🎹🔻🍃📖", "date": "2026-10-29", "surface": "Google AI Overview (expanded)", "auth": "signed out, incognito", "sha256": "9c284bc639b5beb5f2e124db6cf0af4ee8b3095e22d51423073f32ab5cf3408f", "sha256_of": "the registry's 'transcript' string (utf-8)", "sha256_raw": "7316eb803a7108b7dbd819f4330d2aa8320022f94e90b97320854e3b8c946b9b", "words": 219, "cards": 6, "card_rail": [ "1. OECD AI Policy Observatory — OECD AI Incidents Monitor, an evidence base for trustworthy AI", "2. OECD — AI risks and incidents", "3. OECD AI Policy Observatory — Overview and methodology of the AI Incidents and Hazards Monitor - OECD.AI", "4. OECD — Trends in AI incidents and hazards reported by the media | OECD", "5. OECD — Defining AI incidents and related terms (EN)", "6. YouTube·International Association for Safe & Ethical AI — Establishing Effective Serious Incident Reporting Frameworks – Jimmy Farrell ... (video, 29:48)" ], "claims": [ "T1-1 definition: 'a real-time tracking tool developed by the OECD.AI Policy Observatory in collaboration with the Patrick J. McGovern Foundation to monitor actual and potential risks posed by artificial intelligence' (marker: an opaque token, the one shared with the Real-Time Tracking line and the incident definition)", "T1-2 Real-Time Tracking: 'It continuously aggregates and analyzes world news using the Event Registry platform, processing over 150,000 news articles daily to detect AI-related events.'", "T1-3 Evidence Base: 'It provides policymakers, developers, and the public with empirical data on how AI risks materialize globally.'", "T1-4 Thematic Classification: 'Media reports are categorized into 14 distinct risk themes—including synthetic media, cyberattacks and fraud, autonomous vehicles, large language model (LLM) hallucinations, labour market disruptions, and child safety.'", "T1-5 AI Incident: 'An event where the development, use, or malfunction of an AI system results in actual harm (such as injury, health damage, critical infrastructure disruption, or human/fundamental rights violations).'", "T1-6 AI Hazard: 'An event or circumstance where the development or use of an AI system could plausibly lead to an AI incident (a near-harm or precursor event).'", "T1-7 access: 'You can explore live data, interactive filters, and recent tracking reports directly on the OECD AI Incidents Monitor.'", "T1-8 closer: 'Would you like to explore specific risk categories (such as labor market impacts or synthetic media) or look into the common reporting framework guidelines?'" ], "markers": "inline links are opaque google.com/goto tokens; one token marks the definition, the Real-Time Tracking line and the second marker of the incident definition; another marks the OECD.AI Policy Observatory link and the access line; none resolves to a card by itself", "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/T1-oecd-ai-incidents-and-hazards-monitor-aio-20261029.txt" } ] }, "KO": { "title": "The OECD AI Incidents and Hazards Monitor", "genre": "knowledge object, equal case, under EA-NEGONT-02 v0.9 (#1690): the archive's sources and the field's admitted on equal terms, ontology included, in one claim space", "status": "composed 2026-10-29 under v0.9 (brief rebuild/negative-of-the-negative/v09/BRIEF-NEW-ROW-AGGREGATOR.md); not frozen; ledgers unaudited", "sentences": [ { "n": 1, "para": 0, "sense": "what it is", "text": "The OECD AI Incidents and Hazards Monitor (AIM) is an automated monitor of AI harm events in the news, run by OECD.AI, whose census gives its kind as a monitor (MMRS record); its own page calls it an \"Automated media discourse monitor of AI incidents and hazards (Beta)\", which the MMRS record quotes, and the OECD's trends paper of February 2026 describes it as tracking media-reported AI incidents and hazards \"in real-time\" and leveraging the OECD common reporting framework to categorise them consistently.", "claims": [ "MMRS-01", "MMRS-19", "F1", "MMRS-03", "F38" ] }, { "n": 2, "para": 0, "sense": "name and terms", "text": "Its methodology page records that it was previously the AI Incidents Monitor, the name the MMRS record gives beside AIM, and that its use is subject to the OECD's terms and conditions.", "claims": [ "F22", "MMRS-17" ] }, { "n": 3, "para": 0, "sense": "maker", "text": "The methodology page says it \"was initiated and is being developed by the OECD.AI expert group on AI incidents with the support of the Patrick J. McGovern Foundation\"; the OECD's definitions paper dates the group to January 2023, when the OECD formalised it to develop a common AI incident reporting framework and an AI Incidents Monitor, and the MMRS record names OECD.AI as operator, on the methodology page's evidence.", "claims": [ "F11", "F50", "MMRS-18" ] }, { "n": 4, "para": 0, "sense": "purpose", "text": "Its own page says it documents incidents and hazards so that policymakers, AI practitioners and all stakeholders may gain insight into the risks and harms of AI systems, and that over time it will help show risk patterns and a collective understanding of them; the methodology page sets its goal as tracking incidents and hazards \"as defined by the OECD, in real time\" to provide the evidence base for policy discussion, and the OECD's topic page, read by its card's snippet only, says all actors must monitor incidents and hazards and that risk mitigation will require a solid evidence base.", "claims": [ "F2", "F3", "F12", "F10" ] }, { "n": 5, "para": 1, "sense": "definitions", "text": "Its terms are the OECD's: the paper Defining AI incidents and related terms (OECD Artificial Intelligence Papers No. 16, May 2024), written by Karine Perset and Luis Aranda, approved by the Digital Policy Committee on 14 March 2024 and built on a stocktaking of more than 30 frameworks, standards and legal instruments, provides them to inform the common reporting framework and \"its application through the OECD AI Incidents Monitor (AIM)\", and the methodology page and the MMRS record both source the monitor's definitions to it.", "claims": [ "F48", "F49", "F59", "F18", "MMRS-05" ] }, { "n": 6, "para": 1, "sense": "definitions", "text": "An AI incident is \"an event, circumstance or series of events where the development, use or malfunction of one or more AI systems directly or indirectly leads to\" harm to the health of a person or groups, disruption of critical infrastructure, violations of human rights or breaches of law protecting fundamental, labour and intellectual property rights, or harm to property, communities or the environment (the paper; restated on the methodology page), and an AI hazard is such an event that \"could plausibly lead to an AI incident\", near misses included (the paper; the methodology page); the MMRS record gives the two terms in brief, the hazard in the paper's words.", "claims": [ "F51", "F19", "F52", "F20", "F53", "MMRS-04" ] }, { "n": 7, "para": 1, "sense": "definitions", "text": "The paper places reputational harm to individuals, hate speech and mis- and disinformation under (c), as a breach of fundamental rights, and violations of authors' and personal rights there too, with industrial property under (d); the definitions rest on the OECD Recommendation's definition of an AI system (2019, revised 2024), as the methodology page states.", "claims": [ "F56", "F21" ] }, { "n": 8, "para": 1, "sense": "the graded terms", "text": "The paper also grades the terms, serious AI incidents a subset \"intended to include substantial, material incidents\", an AI disaster a serious incident that may test or exceed a society's capacity to cope, and serious AI hazards a subset of hazards, and leaves seriousness context-dependent and \"left for further discussion\"; its governing limit is that the definitions aim at interoperability \"while providing flexibility for jurisdictions to determine the scope of AI incidents and hazards they wish to address\".", "claims": [ "F54", "F55", "F58", "F57" ] }, { "n": 9, "para": 2, "sense": "the unit", "text": "Its unit is the event: an incident or hazard \"can be reported by one or more news articles covering the same event\", its own page notes, its data processing powered by Microsoft Azure using data from Event Registry, and the MMRS record's census gives the unit as an incident or hazard, an event in which an AI system leads, or could plausibly lead, to harm.", "claims": [ "F6", "MMRS-20" ] }, { "n": 10, "para": 2, "sense": "intake", "text": "Its intake is the news: the methodology page says the monitor detects incidents and hazards \"from reputable international news outlets\", with data from Event Registry, a news intelligence platform \"processing over 150 000 news articles daily\", whose events are clusters of articles on one happening tagged with concepts, only events with AI concepts reaching the monitor; the trends paper adds that the data rely on the top 50% of media outlets by Alexa traffic ranking, and the MMRS record gives the same clustering and tagging and reads admission as an event entering when reputable outlets report it.", "claims": [ "F13", "F14", "F39", "MMRS-06", "MMRS-10" ] }, { "n": 11, "para": 3, "sense": "method", "text": "Since November 2024 its classifiers are large language models (methodology page): from Event Registry's event summary, OpenAI's GPT-4o mini filters out events that seem unrelated and GPT-4o reclassifies the rest as incidents or hazards, each article is then classified from its own content and the event takes the most common label among its articles, and items collected before November 2024, classified then \"using traditional machine learning techniques\", are being re-annotated with the same method.", "claims": [ "F25", "F26", "F27", "F34" ] }, { "n": 12, "para": 3, "sense": "method", "text": "Similar events are grouped through the ten nearest summaries by cosine similarity, o3-mini deciding whether they cluster, and the pipeline runs daily on events one to four days old, articles later than that being \"likely to be misclassified as a separate event\" (methodology page).", "claims": [ "F30", "F31" ] }, { "n": 13, "para": 3, "sense": "method", "text": "The MMRS record gives the same pipeline: AI-tagged news events read from Event Registry daily and, since November 2024, large language models that sort them into incidents and hazards and write their summaries (GPT-4o mini filtering, GPT-4o reclassifying, o3-mini writing the title, summary, harm type, severity, stakeholders and country from the top three articles), new items matched to existing events by embedding similarity, its census intake \"automated\" and its cadence daily.", "claims": [ "MMRS-02", "MMRS-07", "MMRS-08", "MMRS-21", "MMRS-09", "MMRS-23" ] }, { "n": 14, "para": 4, "sense": "the record", "text": "Each record's title, summary and metadata are written by a model: the methodology page says the title, summary, harm type, severity, affected stakeholders and country are \"LLM-generated (OpenAI’s o3-mini) from the top three articles of each event\", the articles chosen by their outlets' ranking in their country and the metadata generated once, when the item is added, and on the monitor's own page its summaries and the explanations of its labels are marked \"[AI generated]\", some explanations ruling out further classes by name (\"It is not Complementary Information … nor is it Beneficial Use or Unrelated\").", "claims": [ "F28", "F29", "F8", "F9" ] }, { "n": 15, "para": 4, "sense": "the monitor as machine reception", "text": "The MMRS record draws the consequence: the record does not preserve the original outputs, its titles, summaries and metadata LLM-written from the top articles, so \"the monitor is itself a machine reception of the news: its summaries are compositions, and the articles are their sources\"; the Capture Registry (2026-10-29) sets this reading against the composition layer's, which reads the monitor \"as an aggregator of the news\".", "claims": [ "MMRS-25", "MMRS-12", "CAP-1-03" ] }, { "n": 16, "para": 5, "sense": "count", "text": "Its page listed \"About 18579 incidents & hazards\" on 29 October 2026, the count the MMRS record observed on 28 October; the methodology page gives two daily rates, about 1,000 events analysed a day \"resulting in around 10 AI incidents or hazards\" and, in its table's note, \"on average 30 incidents and hazards per day\".", "claims": [ "F7", "MMRS-22", "F32", "F33" ] }, { "n": 17, "para": 5, "sense": "trend", "text": "Its page says incidents \"seem to be getting more media attention lately\" but have gone down as a share of all AI news, and the trends paper counts 92 a month in 2022 rising to 324 in 2025 while their share of AI news fell from 3.2% to around 2.5%.", "claims": [ "F4", "F40" ] }, { "n": 18, "para": 6, "sense": "limits", "text": "Its limits are stated by its sources: the methodology page recognises that these publicly reported incidents and hazards likely \"only represent a subset of all AI incidents and hazards worldwide\", a starting point for an evidence base, and the trends paper that its findings \"reflect media and public interest rather than comprehensive or systematic tracking\", media capturing \"only what journalists are aware of and deemed newsworthy\", that reliance on media may admit incidents mischaracterised, exaggerated or inaccurately attributed to AI, and that its trends do not measure severity, harm type or the number of stakeholders affected.", "claims": [ "F15", "F36", "F37", "F44" ] }, { "n": 19, "para": 6, "sense": "limits", "text": "Its disclosures say that its information should not be reported as the official views of the OECD or its members, that it is populated with articles from third-party outlets and news aggregators with which the OECD has no affiliation, that the OECD \"does not independently verify\" their accuracy, completeness or validity, and that collection and analysis seek reliability, objectivity and quality \"to the best extent possible\".", "claims": [ "F5", "F23", "F24", "F17" ] }, { "n": 20, "para": 6, "sense": "what its selection leaves out", "text": "What its selection leaves out, in the MMRS record's reading: admission requires harm, an incident or a hazard that could plausibly lead to one, and \"a harm no outlet reports has no event to cluster\".", "claims": [ "MMRS-24", "MMRS-11" ] }, { "n": 21, "para": 6, "sense": "the record kept apart", "text": "The MMRS record keeps itself apart from the entity, each assertion carrying a source and a basis and the lifecycle kept apart from the observation that saw it: the monitor active as read on 28 October 2026, its launch date and its licence unchecked; the trends paper carries its own licence on its rights page, Creative Commons Attribution 4.0.", "claims": [ "MMRS-31", "MMRS-27", "MMRS-26", "F47" ] }, { "n": 22, "para": 7, "sense": "themes", "text": "The fourteen themes are the trends paper's (OECD Artificial Intelligence Papers No. 53, February 2026): it classifies the monitor's media reports into 14 themes, among them synthetic media, child safety, cyberattacks, privacy and health, with three patterns of coverage, increasing, intermittent and declining, by collecting the monitor's \"LLM-generated summaries\", clustering them by k-means into 25 clusters, a number chosen \"based on a qualitative keywords review\" when the statistical tests gave inconsistent results, consolidating them into 14 and assigning each item a single theme.", "claims": [ "F35", "F41", "F42", "F43" ] }, { "n": 23, "para": 7, "sense": "themes", "text": "The Capture Registry (2026-10-29) records that the composition's count of fourteen themes comes from that card, that the methodology page gives no count of themes, and that it did not verify the count.", "claims": [ "CAP-1-06" ] }, { "n": 24, "para": 8, "sense": "place", "text": "The trends paper places the monitor where there are \"few public and comparable sources of information on AI incidents and hazards\", media reports among the proxy indicators that give partial insight, beside the common framework for reporting AI incidents that the Global Partnership on AI approved in early 2025; the methodology page names as next steps an open submission process under that framework and court rulings and supervisory decisions, and the trends paper open submissions and a record reorganised by theme.", "claims": [ "F45", "F16", "F46" ] }, { "n": 25, "para": 8, "sense": "neighbours", "text": "In the MMRS record it is the automated counterpart of the AI Incident Database, \"whose editors resolve each submission\", and with the AIAAIC Repository one of the census's three harm repositories, its neighbours in the harm basin, \"automated and news-derived\" in the MMRS record of the AI Incident Database; the definitions paper thanks that database's founder, Sean McGregor (Responsible AI Collaborative), among the experts who gave feedback.", "claims": [ "MMRS-13", "MMRS-16", "MMRS-29", "MMRS-15", "F60" ] }, { "n": 26, "para": 8, "sense": "holding", "text": "A research study can enter it as an event (MMRS record): the Tow Center's study of AI search citation is filed as an incident, \"a research study entering a harm monitor as an event\", and on its page the item, titled \"AI Search Engines Fabricate News Citations, Undermining Publishers\", carries a machine-written summary in which eight search engines and chatbots \"frequently misquote or fabricate news citations, remixing publisher content without credit\", is labelled an AI incident with harm types economic/property, reputational and public interest, and is built from news articles about the study, the earliest dated 11 March 2025.", "claims": [ "MMRS-14", "MMRS-28", "F62", "F63", "F64" ] }, { "n": 27, "para": 8, "sense": "relation", "text": "The archive sets one of its own instruments apart from the class: \"Media monitors track coverage of events\", while the Encyclotron (#637) tracks the representation of entire fields of scholarship, \"not what was said yesterday but what the retrieval layer thinks the field has been saying for decades\".", "claims": [ "O637-01" ] }, { "n": 28, "para": 9, "sense": "reception", "text": "Its card rail at its name also carries a conference talk admitted by its title, \"Establishing Effective Serious Incident Reporting Frameworks – Jimmy Farrell | IASEAI 2025\".", "claims": [ "F61" ] }, { "n": 29, "para": 9, "sense": "reception", "text": "At its name the Overview gave the monitor's input as its methodology page does, Event Registry and 150,000 articles a day, and a maker close to the page's, and did not say that large language models have classified its events since November 2024 and written their titles and summaries: it composed the monitor \"with its inputs and definitions right and its classifier left out, the part of its method that makes each record a machine composition\" (Capture Registry, 2026-10-29, checked against the MMRS record and the entity's own pages), the record holding \"what the composition drops\", and the MMRS record lists the capture as the composition layer's account of the entity.", "claims": [ "CAP-1-01", "CAP-1-05", "CAP-1-02", "CAP-1-07", "CAP-1-04", "CAP-1-08", "MMRS-30" ] } ], "joins": [ { "act": "order of senses", "sentences": "1–29", "claims": [], "note": "equal case (§4.7): the entity's dependency order, field and archive claims ordered together by genealogy within each sense; neither ledger placed first as a block. What it is, its names, maker and purpose (¶0) precede the terms (¶1) every later sense uses; the unit and intake (¶2) precede the method that classifies them (¶3) and the record the method writes (¶4), from which the archive's reading of the monitor as a machine reception follows; counts and trend (¶5) and limits and what the selection leaves out (¶6) qualify the record; the themes (¶7) are a later analysis of the record's summaries; place, neighbours, the holding and the archive's class distinction (¶8) and reception (¶9) stand last, as relations of the entity already composed." }, { "act": "holding together (two daily rates)", "sentences": [ 16 ], "claims": [ "F7", "MMRS-22", "F32", "F33" ], "note": "P_coh: one page, two figures (about 10, about 30 a day), composed each with its locus; the composer does not rule which holds (open block)" }, { "act": "holding together (machine-written record and its reading)", "sentences": [ 14, 15 ], "claims": [ "F28", "F8", "MMRS-25", "MMRS-12", "CAP-1-03" ], "note": "the field's statements that the record is LLM-written precede the MMRS record's reading, which draws the consequence ('so'); the Capture Registry's contrast of readings follows; no claim raised" }, { "act": "placing the themes after the record", "sentences": [ 22, 23 ], "claims": [ "F35", "F41", "F42", "F43", "CAP-1-06" ], "note": "the themes are composed as the trends paper's analysis because its method (F41) clusters the record's LLM-generated summaries; the capture's unverified count follows" }, { "act": "placing in the open block", "sentences": [], "claims": [ "CAP-1-06", "F10", "F22", "F26", "F31", "F32", "F33", "F34", "F35", "F38", "F41", "F46", "F47", "F50", "F56", "F61", "F62", "F63", "F7", "F9", "MMRS-11", "MMRS-14", "MMRS-22", "MMRS-26" ], "note": "each open question rests on claims composed in the body at their own modality" } ], "plan": { "order": [ "MMRS-01", "MMRS-19", "F1", "MMRS-03", "F38", "F22", "MMRS-17", "F11", "F50", "MMRS-18", "F2", "F3", "F12", "F10", "F48", "F49", "F59", "F18", "MMRS-05", "F51", "F19", "F52", "F20", "F53", "MMRS-04", "F56", "F21", "F54", "F55", "F58", "F57", "F6", "MMRS-20", "F13", "F14", "F39", "MMRS-06", "MMRS-10", "F25", "F26", "F27", "F34", "F30", "F31", "MMRS-02", "MMRS-07", "MMRS-08", "MMRS-21", "MMRS-09", "MMRS-23", "F28", "F29", "F8", "F9", "MMRS-25", "MMRS-12", "CAP-1-03", "F7", "MMRS-22", "F32", "F33", "F4", "F40", "F15", "F36", "F37", "F44", "F5", "F23", "F24", "F17", "MMRS-24", "MMRS-11", "MMRS-31", "MMRS-27", "MMRS-26", "F47", "F35", "F41", "F42", "F43", "CAP-1-06", "F45", "F16", "F46", "MMRS-13", "MMRS-16", "MMRS-29", "MMRS-15", "F60", "MMRS-14", "MMRS-28", "F62", "F63", "F64", "O637-01", "F61", "CAP-1-01", "CAP-1-05", "CAP-1-02", "CAP-1-07", "CAP-1-04", "CAP-1-08", "MMRS-30" ], "senses": { "what it is": [ 1 ], "name and terms": [ 2 ], "maker": [ 3 ], "purpose": [ 4 ], "definitions": [ 5, 6, 7 ], "the graded terms": [ 8 ], "the unit": [ 9 ], "intake": [ 10 ], "method": [ 11, 12, 13 ], "the record": [ 14 ], "the monitor as machine reception": [ 15 ], "count": [ 16 ], "trend": [ 17 ], "limits": [ 18, 19 ], "what its selection leaves out": [ 20 ], "the record kept apart": [ 21 ], "themes": [ 22, 23 ], "place": [ 24 ], "neighbours": [ 25 ], "holding": [ 26 ], "relation": [ 27 ], "reception": [ 28, 29 ] }, "case_by_sense": { "equal (one claim space; nothing translated)": [ "what it is", "name and terms", "maker", "purpose", "definitions", "the graded terms", "the unit", "intake", "method", "the record", "the monitor as machine reception", "count", "trend", "limits", "what its selection leaves out", "the record kept apart", "themes", "place", "neighbours", "holding", "relation", "reception" ] }, "carried": { "archive": 40, "field": 64, "not_composed": {} }, "conflict_log": [ { "rule": "D_pres vs channel", "result": "no cut: the KO carries every field and archive claim; the compression's bound met (body 350 words) with no channel log" }, { "rule": "P_coh", "result": "F32 against F33 (sentence 16; internally_inconsistent in B3); MMRS-18 beside F11 (sentence 3); MMRS-22 beside F7 (sentence 16): each composed with its source" }, { "rule": "N_c", "result": "no claim lowered by standing; the MMRS record's readings (MMRS-10, MMRS-11, MMRS-12, MMRS-13) carried at interpretation, as the record codes them; #637's distinction carried at interpretation" }, { "rule": "C_ex", "result": "senses composed ⊇ senses in the ledgers: 26 lineages, every ledger lineage cited in the KO" } ] } }, "P": { "title": "OECD AI Incidents and Hazards Monitor", "genre": "the compression: AIO's interaction grammar (lede, headed clusters, bolded labels, card rail), composed from the v0.9 plan in one claim space; it faces on the card, with AIO's transcript a record within it", "status": "composed 2026-10-29 from the v0.9 knowledge object; not frozen", "lede": { "text": "is an automated monitor of AI harm events in the news, run by OECD.AI (MMRS record), \"Automated media discourse monitor of AI incidents and hazards (Beta)\" (its page): since November 2024 language models classify its events and write each record (methodology page).", "claims": [ "MMRS-01", "F1", "F26", "F28" ] }, "sections": [ { "icon": "🗂️", "head": "What it is", "items": [ { "label": "Maker:", "text": "\"initiated and is being developed by the OECD.AI expert group on AI incidents\", McGovern Foundation support; formalised January 2023 (OECD); operator OECD.AI (MMRS record).", "claims": [ "F11", "F50", "MMRS-18" ] }, { "label": "Names:", "text": "AIM; previously AI Incidents Monitor (methodology page).", "claims": [ "F22", "MMRS-17" ] }, { "label": "Purpose:", "text": "an evidence base for policy, \"in real time\" (its pages; OECD).", "claims": [ "F12", "F2", "F3", "F10" ] } ] }, { "icon": "📖", "head": "Its terms (OECD, May 2024)", "items": [ { "label": "Incident:", "text": "an AI system's development, use or malfunction that \"directly or indirectly leads to\" harm to health, infrastructure, rights, or property, communities, environment.", "claims": [ "F51", "F19", "F18", "F48", "F49", "F59", "F21", "MMRS-04", "MMRS-05" ] }, { "label": "Hazard:", "text": "an event that \"could plausibly lead to an AI incident\", near misses included.", "claims": [ "F52", "F20", "F53" ] }, { "label": "Graded:", "text": "serious incident, disaster, serious hazard; misinformation, authors' rights under (c); jurisdictions set scope.", "claims": [ "F54", "F55", "F58", "F56", "F57" ] } ] }, { "icon": "⚙️", "head": "How it works (methodology page; MMRS record)", "items": [ { "label": "Intake:", "text": "Event Registry's AI-tagged news events, reputable outlets, top 50% by Alexa rank.", "claims": [ "F13", "F14", "F39", "F6", "MMRS-06", "MMRS-10", "MMRS-20" ] }, { "label": "Classifier:", "text": "GPT-4o mini filters, GPT-4o reclassifies, articles vote; earlier, traditional machine learning.", "claims": [ "F25", "F27", "F34", "MMRS-02", "MMRS-07", "MMRS-21" ] }, { "label": "Record:", "text": "o3-mini writes title, summary, harm, severity, stakeholders, country from top three articles, marked \"[AI generated]\".", "claims": [ "F29", "F8", "F9", "MMRS-25" ] }, { "label": "Cadence:", "text": "daily, events one to four days old; o3-mini clusters similar summaries.", "claims": [ "F31", "F30", "MMRS-08", "MMRS-09", "MMRS-23" ] } ] }, { "icon": "🔍", "head": "A machine reception of the news", "items": [ { "label": "Reading:", "text": "\"its summaries are compositions, and the articles are their sources\" (MMRS record).", "claims": [ "MMRS-12", "CAP-1-03" ] }, { "label": "Unseen:", "text": "\"A harm no outlet reports has no event to cluster\" (MMRS record); \"only what journalists are aware of and deemed newsworthy\" (OECD, 2026).", "claims": [ "MMRS-11", "MMRS-24", "F36", "F15", "F37", "F44" ] }, { "label": "Disclaimers:", "text": "not the OECD's view; articles unverified (methodology page).", "claims": [ "F5", "F23", "F24", "F17" ] } ] }, { "icon": "📊", "head": "Counts and themes", "items": [ { "label": "Listed:", "text": "about 18,579 (its page; MMRS record); 10, or 30, a day (methodology page).", "claims": [ "F7", "MMRS-22", "F32", "F33" ] }, { "label": "Trend:", "text": "92 to 324 a month, 2022–2025, a falling share of AI news (OECD; its page).", "claims": [ "F40", "F4" ] }, { "label": "Themes:", "text": "fourteen, the trends paper's k-means clusters of the monitor's LLM summaries; none on the methodology page (Capture Registry).", "claims": [ "F35", "F41", "F42", "F43", "CAP-1-06" ] } ] }, { "icon": "🧭", "head": "Place", "items": [ { "label": "Neighbours:", "text": "automated counterpart of the editor-resolved AI Incident Database; the AIAAIC (MMRS record).", "claims": [ "MMRS-13", "MMRS-15", "MMRS-16", "MMRS-29", "F60" ] }, { "label": "Holding:", "text": "the Tow Center's AI search citation study, filed as an incident (MMRS record; its page).", "claims": [ "MMRS-14", "MMRS-28", "F62", "F63", "F64" ] }, { "label": "Next:", "text": "open submission under the common reporting framework (methodology page; OECD).", "claims": [ "F45", "F16", "F46" ] }, { "label": "Archive:", "text": "\"Media monitors track coverage of events\" (#637).", "claims": [ "O637-01" ] }, { "label": "Overview:", "text": "inputs and definitions right, \"its classifier left out\" (Capture Registry).", "claims": [ "CAP-1-07", "CAP-1-02", "CAP-1-01", "CAP-1-05" ] } ] } ], "open": [ { "question": "About ten a day or about thirty?", "rests_on": [ "F32", "F33", "F7", "MMRS-22" ], "modality": "documented, both (one page, two figures: internally_inconsistent)", "would_resolve": "the count read on successive days, or the methodology page's own correction" }, { "question": "Did the count move between 28 and 29 October 2026?", "rests_on": [ "F7", "MMRS-22", "F31" ], "modality": "documented (its page, 2026-10-29) and documented (MMRS record, observed 2026-10-28)", "would_resolve": "the count fetched at a known interval against the daily pipeline" }, { "question": "Are the fourteen themes part of the monitor, or the trends paper's analysis of its data?", "rests_on": [ "F35", "F41", "F46", "CAP-1-06" ], "modality": "documented (the paper's method); hypothesis (its next step, a record reorganised by theme); self-description (the Capture Registry's unverified count)", "would_resolve": "the monitor's interface or methodology page offering the themes as a classification of its items" }, { "question": "Three classes or more?", "rests_on": [ "F26", "F9" ], "modality": "documented, both: the methodology page names incidents, hazards and unrelated events; the machine-written explanations also rule out Complementary Information and Beneficial Use", "would_resolve": "a methodology statement of the further labels" }, { "question": "When did it launch, and under what licence are its data?", "rests_on": [ "MMRS-26", "F50", "F34", "F22", "F47" ], "modality": "self-description (the MMRS record's unchecked items); documented (the expert group formalised January 2023; a pre-November-2024 method; the OECD's terms; the trends paper's own CC BY 4.0)", "would_resolve": "the monitor's launch announcement and a stated licence for its data" }, { "question": "Does an answer surface's failure to credit its sources enter only when a study reports it?", "rests_on": [ "MMRS-14", "MMRS-11", "F62", "F63", "F56" ], "modality": "documented (the holding, labelled reputational and public-interest harm); interpretation (the MMRS record); stipulation (authors' rights under (c))", "would_resolve": "items in the monitor whose event is an answer surface's attribution or citation failure reported without a study" }, { "question": "Opacities", "rests_on": [ "F10", "F38", "F61" ], "modality": "B2 read by its card snippet only (HTTP 403); B4's card page 403, the same publication read as its PDF; B6 by its title only", "would_resolve": "the pages fetched without a challenge; the talk's transcript" } ], "rail": [ { "lineage": "what it is: an automated media monitor", "claims": [ "MMRS-01", "MMRS-19", "F1", "MMRS-03", "F38" ], "earliest": "B4" }, { "lineage": "name and terms", "claims": [ "F22", "MMRS-17", "F47" ], "earliest": "B4" }, { "lineage": "maker: the OECD.AI expert group, McGovern support", "claims": [ "F11", "F50", "MMRS-18" ], "earliest": "B5" }, { "lineage": "purpose: an evidence base for policy", "claims": [ "F2", "F3", "F12", "F10" ], "earliest": "B1" }, { "lineage": "definitions: incident and hazard", "claims": [ "F48", "F49", "F59", "F18", "MMRS-05", "F51", "F19", "F52", "F20", "F53", "MMRS-04", "F56", "F21", "F57" ], "earliest": "B5" }, { "lineage": "definitions: the graded terms", "claims": [ "F54", "F55", "F58" ], "earliest": "B5" }, { "lineage": "the unit: the event and its articles", "claims": [ "F6", "MMRS-20" ], "earliest": "B1" }, { "lineage": "intake: Event Registry's news events", "claims": [ "F13", "F14", "F39", "MMRS-06" ], "earliest": "B4" }, { "lineage": "what its selection leaves out", "claims": [ "MMRS-10", "MMRS-24", "MMRS-11" ], "earliest": "MMRS" }, { "lineage": "method: LLM classification since November 2024", "claims": [ "F25", "F26", "F27", "F34", "F30", "MMRS-02", "MMRS-07", "MMRS-08", "MMRS-21" ], "earliest": "B3" }, { "lineage": "cadence: daily, one to four days", "claims": [ "F31", "MMRS-09", "MMRS-23" ], "earliest": "B3" }, { "lineage": "the record: machine-written", "claims": [ "F28", "F29", "F8", "F9", "MMRS-25" ], "earliest": "B3" }, { "lineage": "the monitor as machine reception", "claims": [ "MMRS-12", "CAP-1-03" ], "earliest": "MMRS" }, { "lineage": "count", "claims": [ "F7", "MMRS-22", "F32", "F33" ], "earliest": "B1" }, { "lineage": "trend: counts and shares", "claims": [ "F4", "F40" ], "earliest": "B4" }, { "lineage": "limits: a subset of what happens", "claims": [ "F15", "F36", "F37", "F44" ], "earliest": "B4" }, { "lineage": "limits: the monitor's own disclaimers", "claims": [ "F5", "F23", "F24", "F17" ], "earliest": "B1" }, { "lineage": "the record kept apart from the entity", "claims": [ "MMRS-31", "MMRS-27", "MMRS-26" ], "earliest": "MMRS" }, { "lineage": "themes: the 2026 paper's analysis", "claims": [ "F35", "F41", "F42", "F43", "CAP-1-06" ], "earliest": "B4" }, { "lineage": "place: the common reporting framework and proxies", "claims": [ "F45" ], "earliest": "B4" }, { "lineage": "next steps: open submission, other sources", "claims": [ "F16", "F46" ], "earliest": "B4" }, { "lineage": "neighbours: the AI Incident Database", "claims": [ "MMRS-13", "MMRS-16", "MMRS-29", "MMRS-15", "F60" ], "earliest": "B5" }, { "lineage": "holding: the Tow Center study", "claims": [ "MMRS-14", "MMRS-28", "F62", "F63", "F64" ], "earliest": "B7" }, { "lineage": "relation: media monitors and the archive's instruments", "claims": [ "O637-01" ], "earliest": 637 }, { "lineage": "reception: video", "claims": [ "F61" ], "earliest": "B6" }, { "lineage": "reception: the composition layer", "claims": [ "CAP-1-01", "CAP-1-05", "CAP-1-02", "CAP-1-07", "CAP-1-04", "CAP-1-08", "MMRS-30" ], "earliest": "capture:oecd-ai-incidents-and-hazards-monitor-aio-20261029" } ], "body_words": 350, "entity": "Composed under v0.9, equal case (ρ = extends): KO_B sentences surviving 0 of 11; every field claim carried." }, "P_B": { "title": "OECD AI Incidents and Hazards Monitor", "arm": "B (the field alone)", "genre": "popup — the compression from the field ledger alone (cards, own page; Wikipedia has no article)", "status": "composed 2026-10-29 from the field ledger alone; not frozen", "lede": { "text": "is an \"Automated media discourse monitor of AI incidents and hazards (Beta)\", developed by the OECD.AI expert group on AI incidents with McGovern Foundation support (its pages).", "claims": [ "F1", "F11" ] }, "sections": [ { "icon": "🗂️", "head": "What it is", "items": [ { "label": "Terms:", "text": "incident (harm) and hazard (plausible harm), OECD, May 2024.", "claims": [ "F51", "F52", "F48" ] }, { "label": "Intake:", "text": "Event Registry's AI-tagged news events.", "claims": [ "F13", "F14" ] } ] }, { "icon": "⚙️", "head": "How it works (methodology page; MMRS record)", "items": [ { "label": "Method:", "text": "since November 2024, GPT-4o mini and GPT-4o classify; o3-mini writes each record.", "claims": [ "F25", "F26", "F28" ] }, { "label": "Count:", "text": "about 18,579 listed.", "claims": [ "F7" ] } ] }, { "icon": "⚠️", "head": "Limits", "items": [ { "label": "Subset:", "text": "\"only what journalists are aware of and deemed newsworthy\".", "claims": [ "F15", "F36" ] }, { "label": "Themes:", "text": "fourteen, from a 2026 OECD analysis of its summaries.", "claims": [ "F35", "F41" ] } ] } ], "open": [ { "question": "About ten a day or about thirty?", "rests_on": [ "F32", "F33" ], "modality": "documented, both", "would_resolve": "successive counts" } ], "rail": [ { "lineage": "what it is: an automated media monitor", "claims": [ "F1", "F38" ], "earliest": "B4" }, { "lineage": "purpose: an evidence base for policy", "claims": [ "F2", "F3", "F12", "F10" ], "earliest": "B1" }, { "lineage": "maker: the OECD.AI expert group, McGovern support", "claims": [ "F11", "F50" ], "earliest": "B5" }, { "lineage": "name and terms", "claims": [ "F22", "F47" ], "earliest": "B4" }, { "lineage": "definitions: incident and hazard", "claims": [ "F48", "F49", "F59", "F18", "F19", "F20", "F51", "F52", "F53", "F21", "F56", "F57" ], "earliest": "B5" }, { "lineage": "definitions: the graded terms", "claims": [ "F54", "F55", "F58" ], "earliest": "B5" }, { "lineage": "intake: Event Registry's news events", "claims": [ "F13", "F14", "F39" ], "earliest": "B4" }, { "lineage": "the unit: the event and its articles", "claims": [ "F6" ], "earliest": "B1" }, { "lineage": "method: LLM classification since November 2024", "claims": [ "F25", "F26", "F27", "F30", "F34" ], "earliest": "B3" }, { "lineage": "the record: machine-written", "claims": [ "F28", "F29", "F8", "F9" ], "earliest": "B3" }, { "lineage": "cadence: daily, one to four days", "claims": [ "F31" ], "earliest": "B3" }, { "lineage": "count", "claims": [ "F7", "F32", "F33" ], "earliest": "B1" }, { "lineage": "trend: counts and shares", "claims": [ "F4", "F40" ], "earliest": "B4" }, { "lineage": "themes: the 2026 paper's analysis", "claims": [ "F35", "F41", "F42", "F43" ], "earliest": "B4" }, { "lineage": "limits: a subset of what happens", "claims": [ "F15", "F36", "F37", "F44" ], "earliest": "B4" }, { "lineage": "limits: the monitor's own disclaimers", "claims": [ "F5", "F23", "F24", "F17" ], "earliest": "B1" }, { "lineage": "place: the common reporting framework and proxies", "claims": [ "F45" ], "earliest": "B4" }, { "lineage": "next steps: open submission, other sources", "claims": [ "F16", "F46" ], "earliest": "B4" }, { "lineage": "neighbours: the AI Incident Database", "claims": [ "F60" ], "earliest": "B5" }, { "lineage": "holding: the Tow Center study", "claims": [ "F62", "F63", "F64" ], "earliest": "B7" }, { "lineage": "reception: video", "claims": [ "F61" ], "earliest": "B6" } ], "body_words": 82, "entity": "With no archive claim admitted, the entry is the field's monitor at its own resolution: its two pages, the OECD's two papers, the holding." }, "KO_B": { "title": "The OECD AI Incidents and Hazards Monitor", "arm": "B (the field alone)", "genre": "knowledge object", "status": "composed 2026-10-29; not frozen", "sentences": [ { "n": 1, "para": 0, "text": "The OECD AI Incidents and Hazards Monitor is an \"Automated media discourse monitor of AI incidents and hazards (Beta)\" that documents AI incidents and hazards to help policymakers, AI practitioners and all stakeholders gain insight into the risks and harms of AI systems and that will, over time, help show risk patterns, by its own page.", "claims": [ "F1", "F2", "F3" ] }, { "n": 2, "para": 0, "text": "The OECD.AI expert group on AI incidents, formalised by the OECD in January 2023 to develop a reporting framework and the monitor, initiated it and develops it with the support of the Patrick J. McGovern Foundation, its goal to track incidents and hazards as the OECD defines them, in real time, as an evidence base for policy; it was previously the AI Incidents Monitor, and the OECD's topic page says all actors must monitor incidents and hazards.", "claims": [ "F11", "F50", "F12", "F22", "F10" ] }, { "n": 3, "para": 1, "text": "Its definitions come from the OECD paper of May 2024 (No. 16, by Karine Perset and Luis Aranda, approved on 14 March 2024, from a stocktaking of more than 30 frameworks): an AI incident is an event where the development, use or malfunction of AI systems directly or indirectly leads to harm to health, critical infrastructure, rights, or property, communities or the environment, and an AI hazard is one that could plausibly lead to an incident, near misses included, both resting on the OECD's definition of an AI system.", "claims": [ "F48", "F49", "F59", "F18", "F19", "F20", "F51", "F52", "F53", "F21" ] }, { "n": 4, "para": 1, "text": "The paper places reputational harm, misinformation, authors' rights under (c), grades the terms into serious incidents, disasters and serious hazards, leaves seriousness to context and lets each jurisdiction set its scope.", "claims": [ "F56", "F54", "F55", "F58", "F57" ] }, { "n": 5, "para": 2, "text": "It detects incidents and hazards from reputable international news outlets through Event Registry, which processes over 150,000 articles a day and clusters them into events tagged by concept, only AI-tagged events reaching the monitor, from the top 50% of outlets by Alexa rank; an incident or hazard can be reported by several articles on one event.", "claims": [ "F13", "F14", "F39", "F6" ] }, { "n": 6, "para": 2, "text": "Since November 2024 GPT-4o mini filters events and GPT-4o reclassifies them as incidents, hazards or unrelated, articles are classified by content and the event takes their majority label, o3-mini writes each event's title, summary, harm type, severity, stakeholders and country from its top three articles, chosen by outlet ranking, and groups events by summary similarity; before that, traditional machine learning classified them.", "claims": [ "F25", "F26", "F27", "F28", "F29", "F30", "F34" ] }, { "n": 7, "para": 2, "text": "The pipeline runs daily on events one to four days old; the page marks its summaries and label explanations \"[AI generated]\", naming classes such as Complementary Information and Beneficial Use, and lists about 18,579 incidents and hazards, while the methodology page gives about 10, and elsewhere 30, a day.", "claims": [ "F31", "F8", "F9", "F7", "F32", "F33" ] }, { "n": 8, "para": 3, "text": "Media attention to AI incidents has grown while their share of AI news fell, from 92 to 324 a month and from 3.2% to about 2.5% between 2022 and 2025, and the OECD's trends paper of February 2026, which describes the monitor as tracking media-reported incidents in real time, clusters its LLM-generated summaries into 14 themes, one per item, in three patterns of coverage.", "claims": [ "F4", "F40", "F38", "F35", "F41", "F42", "F43" ] }, { "n": 9, "para": 3, "text": "Its limits are a likely subset of all incidents, media and public interest rather than systematic tracking, what journalists know and deem newsworthy, possible errors of inclusion and trends that do not measure severity; its information is not the OECD's official view, its third-party articles are not independently verified, and reliability is sought to the best extent possible.", "claims": [ "F15", "F36", "F37", "F44", "F5", "F23", "F24", "F17" ] }, { "n": 10, "para": 4, "text": "Few public, comparable sources exist; the common reporting framework approved by GPAI in early 2025, open submissions, court rulings and a record by theme are the next steps; the trends paper is CC BY 4.0; and the AI Incident Database's founder gave feedback on the definitions.", "claims": [ "F45", "F16", "F46", "F47", "F60" ] }, { "n": 11, "para": 4, "text": "Among its items, the Tow Center's study of AI search engines that fabricate news citations is labelled an AI incident with economic, reputational and public-interest harm, built from articles from 11 March 2025; its card rail also carries a talk on serious incident reporting frameworks (IASEAI 2025).", "claims": [ "F62", "F63", "F64", "F61" ] } ] }, "L_B": { "words": 641, "text": "The OECD AI Incidents and Hazards Monitor is an \"Automated media discourse monitor of AI incidents and hazards (Beta)\" that documents AI incidents and hazards to help policymakers, AI practitioners and all stakeholders gain insight into the risks and harms of AI systems and that will, over time, help show risk patterns, by its own page. The OECD.AI expert group on AI incidents, formalised by the OECD in January 2023 to develop a reporting framework and the monitor, initiated it and develops it with the support of the Patrick J. McGovern Foundation, its goal to track incidents and hazards as the OECD defines them, in real time, as an evidence base for policy; it was previously the AI Incidents Monitor, and the OECD's topic page says all actors must monitor incidents and hazards. Its definitions come from the OECD paper of May 2024 (No. 16, by Karine Perset and Luis Aranda, approved on 14 March 2024, from a stocktaking of more than 30 frameworks): an AI incident is an event where the development, use or malfunction of AI systems directly or indirectly leads to harm to health, critical infrastructure, rights, or property, communities or the environment, and an AI hazard is one that could plausibly lead to an incident, near misses included, both resting on the OECD's definition of an AI system. The paper places reputational harm, misinformation, authors' rights under (c), grades the terms into serious incidents, disasters and serious hazards, leaves seriousness to context and lets each jurisdiction set its scope. It detects incidents and hazards from reputable international news outlets through Event Registry, which processes over 150,000 articles a day and clusters them into events tagged by concept, only AI-tagged events reaching the monitor, from the top 50% of outlets by Alexa rank; an incident or hazard can be reported by several articles on one event. Since November 2024 GPT-4o mini filters events and GPT-4o reclassifies them as incidents, hazards or unrelated, articles are classified by content and the event takes their majority label, o3-mini writes each event's title, summary, harm type, severity, stakeholders and country from its top three articles, chosen by outlet ranking, and groups events by summary similarity; before that, traditional machine learning classified them. The pipeline runs daily on events one to four days old; the page marks its summaries and label explanations \"[AI generated]\", naming classes such as Complementary Information and Beneficial Use, and lists about 18,579 incidents and hazards, while the methodology page gives about 10, and elsewhere 30, a day. Media attention to AI incidents has grown while their share of AI news fell, from 92 to 324 a month and from 3.2% to about 2.5% between 2022 and 2025, and the OECD's trends paper of February 2026, which describes the monitor as tracking media-reported incidents in real time, clusters its LLM-generated summaries into 14 themes, one per item, in three patterns of coverage. Its limits are a likely subset of all incidents, media and public interest rather than systematic tracking, what journalists know and deem newsworthy, possible errors of inclusion and trends that do not measure severity; its information is not the OECD's official view, its third-party articles are not independently verified, and reliability is sought to the best extent possible. Few public, comparable sources exist; the common reporting framework approved by GPAI in early 2025, open submissions, court rulings and a record by theme are the next steps; the trends paper is CC BY 4.0; and the AI Incident Database's founder gave feedback on the definitions. Among its items, the Tow Center's study of AI search engines that fabricate news citations is labelled an AI incident with economic, reputational and public-interest harm, built from articles from 11 March 2025; its card rail also carries a talk on serious incident reporting frameworks (IASEAI 2025).", "note": "L_F(B) is realized as the field arm's expansion (objects.KO_B): the field alone, cards and own page; Wikipedia has no article." }, "L_BA": { "words": 2040, "text": "The OECD AI Incidents and Hazards Monitor (AIM) is an automated monitor of AI harm events in the news, run by OECD.AI, whose census gives its kind as a monitor (MMRS record); its own page calls it an \"Automated media discourse monitor of AI incidents and hazards (Beta)\", which the MMRS record quotes, and the OECD's trends paper of February 2026 describes it as tracking media-reported AI incidents and hazards \"in real-time\" and leveraging the OECD common reporting framework to categorise them consistently. Its methodology page records that it was previously the AI Incidents Monitor, the name the MMRS record gives beside AIM, and that its use is subject to the OECD's terms and conditions. The methodology page says it \"was initiated and is being developed by the OECD.AI expert group on AI incidents with the support of the Patrick J. McGovern Foundation\"; the OECD's definitions paper dates the group to January 2023, when the OECD formalised it to develop a common AI incident reporting framework and an AI Incidents Monitor, and the MMRS record names OECD.AI as operator, on the methodology page's evidence. Its own page says it documents incidents and hazards so that policymakers, AI practitioners and all stakeholders may gain insight into the risks and harms of AI systems, and that over time it will help show risk patterns and a collective understanding of them; the methodology page sets its goal as tracking incidents and hazards \"as defined by the OECD, in real time\" to provide the evidence base for policy discussion, and the OECD's topic page, read by its card's snippet only, says all actors must monitor incidents and hazards and that risk mitigation will require a solid evidence base. Its terms are the OECD's: the paper Defining AI incidents and related terms (OECD Artificial Intelligence Papers No. 16, May 2024), written by Karine Perset and Luis Aranda, approved by the Digital Policy Committee on 14 March 2024 and built on a stocktaking of more than 30 frameworks, standards and legal instruments, provides them to inform the common reporting framework and \"its application through the OECD AI Incidents Monitor (AIM)\", and the methodology page and the MMRS record both source the monitor's definitions to it. An AI incident is \"an event, circumstance or series of events where the development, use or malfunction of one or more AI systems directly or indirectly leads to\" harm to the health of a person or groups, disruption of critical infrastructure, violations of human rights or breaches of law protecting fundamental, labour and intellectual property rights, or harm to property, communities or the environment (the paper; restated on the methodology page), and an AI hazard is such an event that \"could plausibly lead to an AI incident\", near misses included (the paper; the methodology page); the MMRS record gives the two terms in brief, the hazard in the paper's words. The paper places reputational harm to individuals, hate speech and mis- and disinformation under (c), as a breach of fundamental rights, and violations of authors' and personal rights there too, with industrial property under (d); the definitions rest on the OECD Recommendation's definition of an AI system (2019, revised 2024), as the methodology page states. The paper also grades the terms, serious AI incidents a subset \"intended to include substantial, material incidents\", an AI disaster a serious incident that may test or exceed a society's capacity to cope, and serious AI hazards a subset of hazards, and leaves seriousness context-dependent and \"left for further discussion\"; its governing limit is that the definitions aim at interoperability \"while providing flexibility for jurisdictions to determine the scope of AI incidents and hazards they wish to address\". Its unit is the event: an incident or hazard \"can be reported by one or more news articles covering the same event\", its own page notes, its data processing powered by Microsoft Azure using data from Event Registry, and the MMRS record's census gives the unit as an incident or hazard, an event in which an AI system leads, or could plausibly lead, to harm. Its intake is the news: the methodology page says the monitor detects incidents and hazards \"from reputable international news outlets\", with data from Event Registry, a news intelligence platform \"processing over 150 000 news articles daily\", whose events are clusters of articles on one happening tagged with concepts, only events with AI concepts reaching the monitor; the trends paper adds that the data rely on the top 50% of media outlets by Alexa traffic ranking, and the MMRS record gives the same clustering and tagging and reads admission as an event entering when reputable outlets report it. Since November 2024 its classifiers are large language models (methodology page): from Event Registry's event summary, OpenAI's GPT-4o mini filters out events that seem unrelated and GPT-4o reclassifies the rest as incidents or hazards, each article is then classified from its own content and the event takes the most common label among its articles, and items collected before November 2024, classified then \"using traditional machine learning techniques\", are being re-annotated with the same method. Similar events are grouped through the ten nearest summaries by cosine similarity, o3-mini deciding whether they cluster, and the pipeline runs daily on events one to four days old, articles later than that being \"likely to be misclassified as a separate event\" (methodology page). The MMRS record gives the same pipeline: AI-tagged news events read from Event Registry daily and, since November 2024, large language models that sort them into incidents and hazards and write their summaries (GPT-4o mini filtering, GPT-4o reclassifying, o3-mini writing the title, summary, harm type, severity, stakeholders and country from the top three articles), new items matched to existing events by embedding similarity, its census intake \"automated\" and its cadence daily. Each record's title, summary and metadata are written by a model: the methodology page says the title, summary, harm type, severity, affected stakeholders and country are \"LLM-generated (OpenAI’s o3-mini) from the top three articles of each event\", the articles chosen by their outlets' ranking in their country and the metadata generated once, when the item is added, and on the monitor's own page its summaries and the explanations of its labels are marked \"[AI generated]\", some explanations ruling out further classes by name (\"It is not Complementary Information … nor is it Beneficial Use or Unrelated\"). The MMRS record draws the consequence: the record does not preserve the original outputs, its titles, summaries and metadata LLM-written from the top articles, so \"the monitor is itself a machine reception of the news: its summaries are compositions, and the articles are their sources\"; the Capture Registry (2026-10-29) sets this reading against the composition layer's, which reads the monitor \"as an aggregator of the news\". Its page listed \"About 18579 incidents & hazards\" on 29 October 2026, the count the MMRS record observed on 28 October; the methodology page gives two daily rates, about 1,000 events analysed a day \"resulting in around 10 AI incidents or hazards\" and, in its table's note, \"on average 30 incidents and hazards per day\". Its page says incidents \"seem to be getting more media attention lately\" but have gone down as a share of all AI news, and the trends paper counts 92 a month in 2022 rising to 324 in 2025 while their share of AI news fell from 3.2% to around 2.5%. Its limits are stated by its sources: the methodology page recognises that these publicly reported incidents and hazards likely \"only represent a subset of all AI incidents and hazards worldwide\", a starting point for an evidence base, and the trends paper that its findings \"reflect media and public interest rather than comprehensive or systematic tracking\", media capturing \"only what journalists are aware of and deemed newsworthy\", that reliance on media may admit incidents mischaracterised, exaggerated or inaccurately attributed to AI, and that its trends do not measure severity, harm type or the number of stakeholders affected. Its disclosures say that its information should not be reported as the official views of the OECD or its members, that it is populated with articles from third-party outlets and news aggregators with which the OECD has no affiliation, that the OECD \"does not independently verify\" their accuracy, completeness or validity, and that collection and analysis seek reliability, objectivity and quality \"to the best extent possible\". What its selection leaves out, in the MMRS record's reading: admission requires harm, an incident or a hazard that could plausibly lead to one, and \"a harm no outlet reports has no event to cluster\". The MMRS record keeps itself apart from the entity, each assertion carrying a source and a basis and the lifecycle kept apart from the observation that saw it: the monitor active as read on 28 October 2026, its launch date and its licence unchecked; the trends paper carries its own licence on its rights page, Creative Commons Attribution 4.0. The fourteen themes are the trends paper's (OECD Artificial Intelligence Papers No. 53, February 2026): it classifies the monitor's media reports into 14 themes, among them synthetic media, child safety, cyberattacks, privacy and health, with three patterns of coverage, increasing, intermittent and declining, by collecting the monitor's \"LLM-generated summaries\", clustering them by k-means into 25 clusters, a number chosen \"based on a qualitative keywords review\" when the statistical tests gave inconsistent results, consolidating them into 14 and assigning each item a single theme. The Capture Registry (2026-10-29) records that the composition's count of fourteen themes comes from that card, that the methodology page gives no count of themes, and that it did not verify the count. The trends paper places the monitor where there are \"few public and comparable sources of information on AI incidents and hazards\", media reports among the proxy indicators that give partial insight, beside the common framework for reporting AI incidents that the Global Partnership on AI approved in early 2025; the methodology page names as next steps an open submission process under that framework and court rulings and supervisory decisions, and the trends paper open submissions and a record reorganised by theme. In the MMRS record it is the automated counterpart of the AI Incident Database, \"whose editors resolve each submission\", and with the AIAAIC Repository one of the census's three harm repositories, its neighbours in the harm basin, \"automated and news-derived\" in the MMRS record of the AI Incident Database; the definitions paper thanks that database's founder, Sean McGregor (Responsible AI Collaborative), among the experts who gave feedback. A research study can enter it as an event (MMRS record): the Tow Center's study of AI search citation is filed as an incident, \"a research study entering a harm monitor as an event\", and on its page the item, titled \"AI Search Engines Fabricate News Citations, Undermining Publishers\", carries a machine-written summary in which eight search engines and chatbots \"frequently misquote or fabricate news citations, remixing publisher content without credit\", is labelled an AI incident with harm types economic/property, reputational and public interest, and is built from news articles about the study, the earliest dated 11 March 2025. The archive sets one of its own instruments apart from the class: \"Media monitors track coverage of events\", while the Encyclotron (#637) tracks the representation of entire fields of scholarship, \"not what was said yesterday but what the retrieval layer thinks the field has been saying for decades\". Its card rail at its name also carries a conference talk admitted by its title, \"Establishing Effective Serious Incident Reporting Frameworks – Jimmy Farrell | IASEAI 2025\". At its name the Overview gave the monitor's input as its methodology page does, Event Registry and 150,000 articles a day, and a maker close to the page's, and did not say that large language models have classified its events since November 2024 and written their titles and summaries: it composed the monitor \"with its inputs and definitions right and its classifier left out, the part of its method that makes each record a machine composition\" (Capture Registry, 2026-10-29, checked against the MMRS record and the entity's own pages), the record holding \"what the composition drops\", and the MMRS record lists the capture as the composition layer's account of the entity.", "rail": [], "note": "L(B ∪ A) in the equal-terms case (v0.9 §0.4), realized as the knowledge object (objects.KO); its rail is the compression's, by lineage." } }, "field": [ { "id": "B1", "card": "OECD AI Policy Observatory — OECD AI Incidents Monitor, an evidence base for trustworthy AI", "fetched": "card; HTTP 200 (2026-10-29T05:06:49Z); B1.html sha256 e1681d645755e7c4", "origin": "card" }, { "id": "B2", "card": "OECD — AI risks and incidents", "fetched": "card; HTTP 403 (2026-10-29T05:08:47Z), unreachable; card snippet stands; B2.html sha256 54fda3deeb29a2f2", "origin": "card" }, { "id": "B3", "card": "OECD AI Policy Observatory — Overview and methodology of the AI Incidents and Hazards Monitor - OECD.AI", "fetched": "card; HTTP 200 (2026-10-29T05:06:50Z); B3.html sha256 26400e79e67de5ee", "origin": "card" }, { "id": "B4", "card": "OECD — Trends in AI incidents and hazards reported by the media | OECD", "fetched": "card; HTTP 200 (2026-10-29T05:08:47Z); B4.pdf sha256 aab2727da192624f", "origin": "card" }, { "id": "B5", "card": "OECD — Defining AI incidents and related terms (EN)", "fetched": "card; HTTP 200 (2026-10-29T05:08:48Z); B5.pdf sha256 9c46e00f1b3ff6de", "origin": "card" }, { "id": "B6", "card": "YouTube·International Association for Safe & Ethical AI — Establishing Effective Serious Incident Reporting Frameworks – Jimmy Farrell", "fetched": "card; HTTP 200 (2026-10-29T05:09:49Z); B6.json sha256 5a063beb92dbc849", "origin": "card" }, { "id": "B7", "card": "OECD AI Policy Observatory — AI Search Engines Fabricate News Citations, Undermining Publishers (AIM item 2025-03-11-8537)", "fetched": "own page; HTTP 200 (2026-10-29T05:08:49Z); B7.html sha256 9edcbeb65fe222f4", "origin": "own_page" }, { "id": "W1", "card": "Wikipedia — no article (the monitor, its former name, OECD.AI, the OECD AI Policy Observatory)", "fetched": "wikipedia; raw wikitext HTTP 404 for seven titles; rendered page 404; API: titles missing; Special:Search no results", "origin": "wikipedia" }, { "id": "W2", "card": "Wikipedia — OECD (no mention of the monitor)", "fetched": "wikipedia; raw wikitext and rendered page HTTP 200 (2026-10-29T05:10:57Z); W2.wiki sha256 03ca69ee493875ab", "origin": "wikipedia" } ], "field_claims": { "F1": { "source": "B1", "origin": "card", "locus": "header (subtitle under 'AIM: AI Incidents and Hazards Monitor')", "quote": "Automated media discourse monitor of AI incidents and hazards (Beta)", "claim": "the monitor's self-description: an automated media discourse monitor of AI incidents and hazards, in beta", "modality": "stipulation", "kind": "definition", "kernel": true, "const": true, "lineage": "what it is: an automated media monitor", "date": "2026-10-29 (page undated; as fetched)", "quote_ok": true, "checked_against": "saved text B1.txt (sha256 74cf4b7530591ace…), whitespace-normalised" }, "F2": { "source": "B1", "origin": "card", "locus": "introduction ¶1 s2", "quote": "The OECD AI Incidents and Hazards Monitor (AIM) documents AI incidents and hazards to help policymakers, AI practitioners, and all stakeholders worldwide gain valuable insights into the risks and harms of AI systems.", "claim": "its purpose: documenting incidents and hazards for policymakers, practitioners and stakeholders", "modality": "stipulation", "kind": "function", "kernel": false, "const": false, "lineage": "purpose: an evidence base for policy", "date": "2026-10-29 (page undated; as fetched)", "quote_ok": true, "checked_against": "saved text B1.txt (sha256 74cf4b7530591ace…), whitespace-normalised" }, "F3": { "source": "B1", "origin": "card", "locus": "introduction ¶1 s3", "quote": "Over time, AIM will help to show risk patterns and establish a collective understanding of AI incidents and hazards and their multifaceted nature, serving as an important tool for trustworthy AI.", "claim": "its expected use: to show risk patterns over time and a collective understanding of incidents and hazards", "modality": "hypothesis", "kind": "prediction", "kernel": false, "const": false, "lineage": "purpose: an evidence base for policy", "date": "2026-10-29 (page undated; as fetched)", "quote_ok": true, "checked_against": "saved text B1.txt (sha256 74cf4b7530591ace…), whitespace-normalised" }, "F4": { "source": "B1", "origin": "card", "locus": "introduction ¶1 s4", "quote": "AI incidents seem to be getting more media attention lately, but they've actually gone down as a share of all AI news (see chart below!).", "claim": "the page's headline trend: more attention, a falling share of all AI news", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "trend: counts and shares", "date": "2026-10-29 (page undated; as fetched)", "quote_ok": true, "checked_against": "saved text B1.txt (sha256 74cf4b7530591ace…), whitespace-normalised" }, "F5": { "source": "B1", "origin": "card", "locus": "introduction ¶2", "quote": "The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.", "claim": "a limit on its standing: the monitor's information is not the OECD's official view", "modality": "self-description", "kind": "limit", "kernel": true, "const": false, "lineage": "limits: the monitor's own disclaimers", "date": "2026-10-29 (page undated; as fetched)", "quote_ok": true, "checked_against": "saved text B1.txt (sha256 74cf4b7530591ace…), whitespace-normalised" }, "F6": { "source": "B1", "origin": "card", "locus": "chart block, note", "quote": "Note: An AI incident or hazard can be reported by one or more news articles covering the same event. Data processing powered by Microsoft Azure using data from Event Registry.", "claim": "the unit is the event, reported by one or more articles; processing on Microsoft Azure with Event Registry data", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "the unit: the event and its articles", "date": "2026-10-29 (page undated; as fetched)", "quote_ok": true, "checked_against": "saved text B1.txt (sha256 74cf4b7530591ace…), whitespace-normalised" }, "F7": { "source": "B1", "origin": "card", "locus": "results line", "quote": "Results: About 18579 incidents & hazards", "claim": "the count on 2026-10-29: about 18,579 incidents and hazards", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "count", "date": "2026-10-29 (page undated; as fetched)", "quote_ok": true, "checked_against": "saved text B1.txt (sha256 74cf4b7530591ace…), whitespace-normalised" }, "F8": { "source": "B1", "origin": "card", "locus": "result list, the first item's explanation (as fetched 2026-10-29)", "quote": "Why's our monitor labelling this an incident or hazard? … Therefore, the event is best classified as an AI Hazard due to the credible risk of AI exacerbating economic disparities and job losses in the future.[AI generated]", "claim": "each listed item carries an explanation of its label, marked '[AI generated]', as are its summaries", "modality": "documented", "kind": "fact", "kernel": true, "const": false, "lineage": "the record: machine-written", "date": "2026-10-29 (page undated; as fetched)", "quote_ok": true, "checked_against": "saved text B1.txt (sha256 74cf4b7530591ace…), whitespace-normalised (multi-part quote joined by ' … ', each part checked)" }, "F9": { "source": "B1", "origin": "card", "locus": "result list, an item's explanation (as fetched 2026-10-29)", "quote": "It is not Complementary Information because the article is not about responses or updates to a past incident, nor is it Beneficial Use or Unrelated.[AI generated]", "claim": "the machine-written explanations rule out further classes by name: Complementary Information, Beneficial Use, Unrelated", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "the record: machine-written", "date": "2026-10-29 (page undated; as fetched)", "quote_ok": true, "checked_against": "saved text B1.txt (sha256 74cf4b7530591ace…), whitespace-normalised" }, "F10": { "source": "B2", "origin": "card", "locus": "card snippet (page HTTP 403)", "quote": "All actors must monitor AI incidents and hazards to mitigate risks As AI adoption continues to grow, successful risk mitigation will require a solid evidence ba...", "claim": "the OECD's topic page: all actors must monitor incidents and hazards; risk mitigation needs an evidence base", "modality": "interpretation", "kind": "interpretation", "kernel": true, "const": false, "lineage": "purpose: an evidence base for policy", "date": "2026-10-29 (card snippet; page unreachable)", "quote_ok": true, "checked_against": "card snippet (the page answered 403)" }, "F11": { "source": "B3", "origin": "card", "locus": "Overview ¶1 s1", "quote": "The AI Incidents and Hazards Monitor (AIM) was initiated and is being developed by the OECD.AI expert group on AI incidents with the support of the Patrick J. McGovern Foundation.", "claim": "its maker: initiated and developed by the OECD.AI expert group on AI incidents, supported by the Patrick J. McGovern Foundation", "modality": "documented", "kind": "attribution", "kernel": false, "const": false, "lineage": "maker: the OECD.AI expert group, McGovern support", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F12": { "source": "B3", "origin": "card", "locus": "Overview ¶1 s2", "quote": "The goal of the AIM is to track actual AI incidents and hazards, as defined by the OECD, in real time and provide the evidence-base to inform related AI policy discussions.", "claim": "its goal: to track incidents and hazards as the OECD defines them, in real time, as an evidence base for policy", "modality": "stipulation", "kind": "function", "kernel": false, "const": false, "lineage": "purpose: an evidence base for policy", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F13": { "source": "B3", "origin": "card", "locus": "Overview ¶2 s1–2", "quote": "The AIM detects AI incidents and hazards from reputable international news outlets. The data is provided by Event Registry (ER), a news intelligence platform that monitors and aggregates world news by processing over 150 000 news articles daily.", "claim": "intake: reputable international news outlets; data from Event Registry, which processes over 150,000 articles a day", "modality": "documented", "kind": "function", "kernel": false, "const": false, "lineage": "intake: Event Registry's news events", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F14": { "source": "B3", "origin": "card", "locus": "Overview ¶2 s3–5", "quote": "Event Registry detects events, which are clusters of news articles reporting on the same happening. These events are tagged with related concepts. The events provided by Event Registry to AIM are limited to those that include AI concepts.", "claim": "Event Registry clusters articles into events and tags them; only events with AI concepts reach the monitor", "modality": "documented", "kind": "function", "kernel": false, "const": false, "lineage": "intake: Event Registry's news events", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F15": { "source": "B3", "origin": "card", "locus": "Overview ¶3", "quote": "While recognising the likelihood that these incidents and hazards only represent a subset of all AI incidents and hazards worldwide, these publicly reported incidents and hazards nonetheless provide a useful starting point for building the evidence base.", "claim": "its stated limit: the publicly reported incidents and hazards are likely a subset of all, a starting point for the evidence base", "modality": "self-description", "kind": "limit", "kernel": true, "const": false, "lineage": "limits: a subset of what happens", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F16": { "source": "B3", "origin": "card", "locus": "Overview ¶4", "quote": "An open submission process following the common reporting framework for AI incidents is currently being developed to complement AI incidents and hazards from news articles. Other next steps include complementing incidents and hazards with court rulings and decisions by supervisory authorities wherever they exist.", "claim": "next steps: an open submission process under the common reporting framework; court rulings and supervisory decisions", "modality": "hypothesis", "kind": "prediction", "kernel": false, "const": false, "lineage": "next steps: open submission, other sources", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F17": { "source": "B3", "origin": "card", "locus": "Overview ¶5", "quote": "The data collection and analysis for the AIM is done to ensure, to the best extent possible, the reliability, objectivity and quality of the information for AI incidents and hazards.", "claim": "its stated aim for the data: reliability, objectivity and quality to the best extent possible", "modality": "self-description", "kind": "limit", "kernel": false, "const": false, "lineage": "limits: the monitor's own disclaimers", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F18": { "source": "B3", "origin": "card", "locus": "Definitions ¶1", "quote": "Published in May 2024, the paper Defining AI incidents and related terms, defines an event where the development or use of an AI system results in actual harm as an “AI incident”, while an event where the development or use of an AI system is potentially harmful is termed an “AI hazard”.", "claim": "the methodology page sources its terms to the OECD paper of May 2024: actual harm an incident, potential harm a hazard", "modality": "attributed", "kind": "attribution", "kernel": false, "const": false, "lineage": "definitions: incident and hazard", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F19": { "source": "B3", "origin": "card", "locus": "Definitions: AI incident", "quote": "An AI incident is an event, circumstance or series of events where the development, use or malfunction of one or more AI systems directly or indirectly leads to any of the following harms:", "claim": "the methodology page restates the paper's definition of an AI incident, with its four harms (a)–(d)", "modality": "stipulation", "kind": "definition", "kernel": false, "const": false, "lineage": "definitions: incident and hazard", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F20": { "source": "B3", "origin": "card", "locus": "Definitions: AI hazard", "quote": "An AI hazard is an event, circumstance or series of events where the development, use or malfunction of one or more AI systems could plausibly lead to an AI incident, i.e., any of the following harms:", "claim": "the methodology page restates the paper's definition of an AI hazard", "modality": "stipulation", "kind": "definition", "kernel": false, "const": false, "lineage": "definitions: incident and hazard", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F21": { "source": "B3", "origin": "card", "locus": "Definitions: AI system", "quote": "These definitions were built based on the definition of an AI system as described in the OECD recommendation on AI of 2019 and revised in 2024. An AI system is a machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments.", "claim": "the definitions rest on the OECD Recommendation's definition of an AI system (2019, revised 2024)", "modality": "stipulation", "kind": "definition", "kernel": false, "const": false, "lineage": "definitions: incident and hazard", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F22": { "source": "B3", "origin": "card", "locus": "Information Transparency Disclosures: Background", "quote": "Your use of the OECD AI incidents and hazards monitor (previously AI Incidents Monitor) (“AIM”) is subject to the terms and conditions found at www.oecd.org/termsandconditions.", "claim": "its former name, AI Incidents Monitor; its use governed by the OECD's terms and conditions", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "name and terms", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F23": { "source": "B3", "origin": "card", "locus": "Information Transparency Disclosures: Third-Party Information", "quote": "the AIM is populated with news articles from various third-party outlets and news aggregators with which the OECD has no affiliation.", "claim": "its contents are third-party news articles from outlets and aggregators the OECD is not affiliated with", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "limits: the monitor's own disclaimers", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F24": { "source": "B3", "origin": "card", "locus": "Information Transparency Disclosures: Errors and Omissions", "quote": "The OECD cannot guarantee and does not independently verify the accuracy, completeness, or validity of third-party information provided in the AIM.", "claim": "the OECD does not independently verify the third-party information in the monitor", "modality": "self-description", "kind": "limit", "kernel": false, "const": false, "lineage": "limits: the monitor's own disclaimers", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F25": { "source": "B3", "origin": "card", "locus": "Methodology ¶1–2", "quote": "Since November 2024, incidents and hazards are identified and added to the AIM through the following steps. Historical incidents and hazards, collected prior to November 2024, are currently being retrospectively annotated using the same methodology to ensure consistency.", "claim": "the present method dates from November 2024; earlier items are being re-annotated with it", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "method: LLM classification since November 2024", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F26": { "source": "B3", "origin": "card", "locus": "Methodology step 2", "quote": "LLMs are used to classify AI-related events into three categories: AI incidents, AI hazards or unrelated events, as defined above. This classification is based on the event summary provided from Event Registry. The classification follows a two-step process: first, a smaller model (OpenAI’s GPT-4o mini) filters out events that seem unrelated to AI incidents and hazards; next, a larger model (OpenAI’s GPT-4o) reclassifies the remaining events to confirm whether they are incidents or hazards.", "claim": "classification by LLMs into incidents, hazards or unrelated, from Event Registry's summary: GPT-4o mini filters, GPT-4o reclassifies", "modality": "documented", "kind": "function", "kernel": true, "const": true, "lineage": "method: LLM classification since November 2024", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F27": { "source": "B3", "origin": "card", "locus": "Methodology step 3", "quote": "Individual news articles are classified into incidents and hazards using LLMs, following the same process as for events. In this case, however, article content is used instead of event summaries to enhance reliability. … If some articles are classified differently, the event will adopt the most common classification among them.", "claim": "articles are classified by LLMs from their content; the event takes the most common label among its articles", "modality": "documented", "kind": "function", "kernel": false, "const": false, "lineage": "method: LLM classification since November 2024", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised (multi-part quote joined by ' … ', each part checked)" }, "F28": { "source": "B3", "origin": "card", "locus": "Methodology step 5", "quote": "Each event is enriched with relevant metadata, including an event title, summary, the harm type, severity, affected stakeholders, and the country where it occurred, among others. This metadata is LLM-generated (OpenAI’s o3-mini) from the top three articles of each event, selected from different news outlets.", "claim": "each record's title, summary, harm type, severity, stakeholders and country are written by o3-mini from the top three articles", "modality": "documented", "kind": "function", "kernel": true, "const": true, "lineage": "the record: machine-written", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F29": { "source": "B3", "origin": "card", "locus": "Methodology step 5, continued", "quote": "Articles are ranked based on their outlets’ ranking in their respective country. This approach ensures a balance between the popularity of the outlets and fair geographical representation. Currently, the summary and metadata are generated at the time the incident or hazard is added to the AIM.", "claim": "the top articles are chosen by their outlets' national ranking; summary and metadata are generated once, when the item is added", "modality": "documented", "kind": "function", "kernel": false, "const": false, "lineage": "the record: machine-written", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F30": { "source": "B3", "origin": "card", "locus": "Methodology step 6", "quote": "Similar events are grouped by analysing their summaries and identifying related events in AIM using cosine similarity. For each new incident or hazard, the top 10 closest summaries (i.e., “embeddings”) from existing events in AIM are identified. An LLM (OpenAI’s o3-mini) is then used to assess whether any of these similar events can be clustered together.", "claim": "grouping: the ten nearest summaries by cosine similarity, then o3-mini decides whether to cluster", "modality": "documented", "kind": "function", "kernel": false, "const": false, "lineage": "method: LLM classification since November 2024", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F31": { "source": "B3", "origin": "card", "locus": "Update frequency ¶1–2", "quote": "The pipeline runs daily, processing events from one to four days ago. … Articles appearing after this timeframe are likely to be misclassified as a separate event.", "claim": "daily, on events one to four days old; later articles are likely misclassified as a separate event", "modality": "documented", "kind": "limit", "kernel": false, "const": false, "lineage": "cadence: daily, one to four days", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised (multi-part quote joined by ' … ', each part checked)" }, "F32": { "source": "B3", "origin": "card", "locus": "Summary table, Identification and classification", "quote": "On average, about 1 000 AI events are analysed daily, resulting in around 10 AI incidents or hazards.", "claim": "about 1,000 events analysed a day yield about 10 incidents or hazards", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "count", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised", "marks": [ "internally_inconsistent (B3 states the daily yield two ways: about 10 (F32) and about 30 (F33), both loci cited)" ] }, "F33": { "source": "B3", "origin": "card", "locus": "Summary table, note", "quote": "the AIM detects on average 30 incidents and hazards per day.", "claim": "the AIM detects about 30 incidents and hazards a day (the same page gives about 10 above)", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "count", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised", "marks": [ "internally_inconsistent (B3: about 30 (F33) against about 10 (F32))" ] }, "F34": { "source": "B3", "origin": "card", "locus": "Methodology, last ¶", "quote": "Prior to November 2024, AI incidents and hazards were classified and included on the AIM using traditional machine learning techniques.", "claim": "before November 2024 the monitor classified with traditional machine learning", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "method: LLM classification since November 2024", "date": "2026-10-29 (page undated; as fetched; describes the method 'since November 2024')", "quote_ok": true, "checked_against": "saved text B3.txt (sha256 edfcba595f3fb6b8…), whitespace-normalised" }, "F35": { "source": "B4", "origin": "card", "locus": "Abstract", "quote": "This paper analyses live data from the OECD.AI Incidents and Hazards Monitor (AIM), classifying media reports into 14 themes, including synthetic media, child safety, cyberattacks, privacy, and health. It identifies three distinct patterns of media coverage – increasing, intermittent, and declining – and traces how media attention shifts across themes over time.", "claim": "the 2026 paper analyses the monitor's data, classifying its media reports into 14 themes, with three patterns of coverage", "modality": "documented", "kind": "fact", "kernel": true, "const": false, "lineage": "themes: the 2026 paper's analysis", "date": "2026-02 (OECD Artificial Intelligence Papers No. 53; PDF created 2026-02-09)", "quote_ok": true, "checked_against": "saved text B4.pdf.txt (sha256 ec4d42002736b023…), whitespace-normalised" }, "F36": { "source": "B4", "origin": "card", "locus": "Executive summary ¶5", "quote": "The findings reflect media and public interest rather than comprehensive or systematic tracking of AI incidents and hazards. Importantly, media reports capture only what journalists are aware of and deemed newsworthy.", "claim": "the paper's governing limit: its findings reflect media and public interest, not systematic tracking; media capture only what journalists know and deem newsworthy", "modality": "self-description", "kind": "limit", "kernel": true, "const": false, "lineage": "limits: a subset of what happens", "date": "2026-02 (OECD Artificial Intelligence Papers No. 53; PDF created 2026-02-09)", "quote_ok": true, "checked_against": "saved text B4.pdf.txt (sha256 ec4d42002736b023…), whitespace-normalised" }, "F37": { "source": "B4", "origin": "card", "locus": "Executive summary ¶6", "quote": "Conversely, reliance on media sources may introduce errors of inclusion, as some reported incidents may be mischaracterised, exaggerated, or inaccurately attributed to AI due to limited background or technical understanding.", "claim": "reliance on media may admit errors: incidents mischaracterised, exaggerated or wrongly attributed to AI", "modality": "self-description", "kind": "limit", "kernel": false, "const": false, "lineage": "limits: a subset of what happens", "date": "2026-02 (OECD Artificial Intelligence Papers No. 53; PDF created 2026-02-09)", "quote_ok": true, "checked_against": "saved text B4.pdf.txt (sha256 ec4d42002736b023…), whitespace-normalised" }, "F38": { "source": "B4", "origin": "card", "locus": "Executive summary ¶7", "quote": "The OECD.AI Incidents and Hazards Monitor (AIM) tracks these media-reported AI incidents and hazards in real-time and leverages the OECD common reporting framework to categorise them consistently.", "claim": "the paper's account of the monitor: tracking media-reported incidents and hazards in real time, categorised by the common reporting framework", "modality": "documented", "kind": "definition", "kernel": false, "const": false, "lineage": "what it is: an automated media monitor", "date": "2026-02 (OECD Artificial Intelligence Papers No. 53; PDF created 2026-02-09)", "quote_ok": true, "checked_against": "saved text B4.pdf.txt (sha256 ec4d42002736b023…), whitespace-normalised" }, "F39": { "source": "B4", "origin": "card", "locus": "§1 Introduction ¶6", "quote": "Currently the data rely on reports from the top 50% of media outlets, based on the Alexa traffic rankings (OECD.AI, 2025[6]). These reports are aggregated by Event Registry, a news intelligence platform which monitors over 150 000 news articles daily.", "claim": "the data rely on the top 50% of outlets by Alexa traffic rank, aggregated by Event Registry", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "intake: Event Registry's news events", "date": "2026-02 (OECD Artificial Intelligence Papers No. 53; PDF created 2026-02-09)", "quote_ok": true, "checked_against": "saved text B4.pdf.txt (sha256 ec4d42002736b023…), whitespace-normalised" }, "F40": { "source": "B4", "origin": "card", "locus": "§1 Introduction ¶8", "quote": "Between 2022 and 2025, the number of AI incidents and hazards reported by the media rose steadily, from 92 to 324 per month on average (Figure 1.1, Panel A). However, their share relative to total AI news declined, from 3.2% in 2022 to around 2.5% in 2025 (Figure 1.1, Panel B).", "claim": "92 to 324 a month between 2022 and 2025; their share of AI news down from 3.2% to about 2.5%", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "trend: counts and shares", "date": "2026-02 (OECD Artificial Intelligence Papers No. 53; PDF created 2026-02-09)", "quote_ok": true, "checked_against": "saved text B4.pdf.txt (sha256 ec4d42002736b023…), whitespace-normalised" }, "F41": { "source": "B4", "origin": "card", "locus": "Annex A, Identifying key thematic clusters ¶1", "quote": "LLM-generated summaries of AI incidents and hazards were collected and transformed into a semantic representation capturing the underlying concepts. Using a k-means clustering algorithm, these reports were classified into 25 thematic clusters each capturing semantically similar risk characteristics. … This resulted in 14 distinct themes.", "claim": "the 14 themes were made by clustering the monitor's LLM-generated summaries (k-means, 25 clusters, consolidated to 14)", "modality": "documented", "kind": "function", "kernel": false, "const": false, "lineage": "themes: the 2026 paper's analysis", "date": "2026-02 (OECD Artificial Intelligence Papers No. 53; PDF created 2026-02-09)", "quote_ok": true, "checked_against": "saved text B4.pdf.txt (sha256 ec4d42002736b023…), whitespace-normalised (multi-part quote joined by ' … ', each part checked)" }, "F42": { "source": "B4", "origin": "card", "locus": "Annex A, Clustering ¶2", "quote": "Since these methods gave inconsistent results, the authors chose 25 clusters based on a qualitative keywords review.", "claim": "the number of clusters was chosen by a qualitative keyword review after the statistical tests disagreed", "modality": "self-description", "kind": "limit", "kernel": false, "const": false, "lineage": "themes: the 2026 paper's analysis", "date": "2026-02 (OECD Artificial Intelligence Papers No. 53; PDF created 2026-02-09)", "quote_ok": true, "checked_against": "saved text B4.pdf.txt (sha256 ec4d42002736b023…), whitespace-normalised" }, "F43": { "source": "B4", "origin": "card", "locus": "§2 ¶2", "quote": "To maintain analytical clarity and avoid double counting, each media-reported incident or hazard is assigned to the single most relevant theme based on the available information.", "claim": "one theme per item", "modality": "stipulation", "kind": "function", "kernel": false, "const": false, "lineage": "themes: the 2026 paper's analysis", "date": "2026-02 (OECD Artificial Intelligence Papers No. 53; PDF created 2026-02-09)", "quote_ok": true, "checked_against": "saved text B4.pdf.txt (sha256 ec4d42002736b023…), whitespace-normalised" }, "F44": { "source": "B4", "origin": "card", "locus": "§2, after Figure 2.1", "quote": "These trends do not measure how severe incidents are, the types of harm involved, or the number of affected stakeholders, and should not be read as indicators of these factors.", "claim": "the trends measure attention, not severity, harm type or the number affected", "modality": "self-description", "kind": "limit", "kernel": false, "const": false, "lineage": "limits: a subset of what happens", "date": "2026-02 (OECD Artificial Intelligence Papers No. 53; PDF created 2026-02-09)", "quote_ok": true, "checked_against": "saved text B4.pdf.txt (sha256 ec4d42002736b023…), whitespace-normalised" }, "F45": { "source": "B4", "origin": "card", "locus": "Executive summary ¶2–4", "quote": "In early 2025, the members of the Global Partnership on AI (GPAI) approved a common framework for reporting AI incidents, designed as an international data-sharing template for public and private organisations across jurisdictions and sectors … At present, there are few public and comparable sources of information on AI incidents and hazards. … Proxy indicators – such as court rulings on AI-related cases or media reports – can offer partial insights into where and what types of incidents are occurring.", "claim": "the common reporting framework (GPAI, early 2025); few public, comparable sources; media reports as a proxy", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "place: the common reporting framework and proxies", "date": "2026-02 (OECD Artificial Intelligence Papers No. 53; PDF created 2026-02-09)", "quote_ok": true, "checked_against": "saved text B4.pdf.txt (sha256 ec4d42002736b023…), whitespace-normalised (multi-part quote joined by ' … ', each part checked)" }, "F46": { "source": "B4", "origin": "card", "locus": "§6 Conclusion ¶4", "quote": "Next steps could include allowing open submissions to complement media-based monitoring and reorganising incidents and hazards visually by theme to support deeper analysis.", "claim": "next steps: open submissions; the record reorganised by theme", "modality": "hypothesis", "kind": "prediction", "kernel": false, "const": false, "lineage": "next steps: open submission, other sources", "date": "2026-02 (OECD Artificial Intelligence Papers No. 53; PDF created 2026-02-09)", "quote_ok": true, "checked_against": "saved text B4.pdf.txt (sha256 ec4d42002736b023…), whitespace-normalised" }, "F47": { "source": "B4", "origin": "card", "locus": "rights page", "quote": "Attribution 4.0 International (CC BY 4.0)", "claim": "the paper (not the monitor's data) is licensed CC BY 4.0", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "name and terms", "date": "2026-02 (OECD Artificial Intelligence Papers No. 53; PDF created 2026-02-09)", "quote_ok": true, "checked_against": "saved text B4.pdf.txt (sha256 ec4d42002736b023…), whitespace-normalised" }, "F48": { "source": "B5", "origin": "card", "locus": "cover; Foreword ¶1", "quote": "May 2024 No. 16 … This report provides a definition of AI incident and related terminology. These definitions aim to clarify what constitutes an AI incident, an AI hazard, and associated terminology without being overly prescriptive. This document will inform the development of a common AI incidents reporting framework and its application through the OECD AI Incidents Monitor (AIM).", "claim": "the definitions paper (OECD AI Papers No. 16, May 2024): definitions of AI incident and related terms, not overly prescriptive, for the reporting framework and the AIM", "modality": "stipulation", "kind": "definition", "kernel": false, "const": false, "lineage": "definitions: incident and hazard", "date": "2024-05 (OECD Artificial Intelligence Papers No. 16; approved by the DPC 2024-03-14)", "quote_ok": true, "checked_against": "saved text B5.pdf.txt (sha256 97db457717a855e2…), whitespace-normalised (multi-part quote joined by ' … ', each part checked)" }, "F49": { "source": "B5", "origin": "card", "locus": "Foreword ¶3–4", "quote": "The report was written by Karine Perset and Luis Aranda under the supervision of Audrey Plonk, Deputy Director of the OECD Science Technology and Innovation Directorate. … This paper was approved and declassified by written procedure by the OECD Digital Policy Committee (DPC) on 14 March 2024 and prepared for publication by the OECD Secretariat.", "claim": "authors Karine Perset and Luis Aranda; approved by the Digital Policy Committee on 14 March 2024", "modality": "documented", "kind": "attribution", "kernel": false, "const": false, "lineage": "definitions: incident and hazard", "date": "2024-05 (OECD Artificial Intelligence Papers No. 16; approved by the DPC 2024-03-14)", "quote_ok": true, "checked_against": "saved text B5.pdf.txt (sha256 97db457717a855e2…), whitespace-normalised (multi-part quote joined by ' … ', each part checked)" }, "F50": { "source": "B5", "origin": "card", "locus": "Executive summary ¶1", "quote": "In January 2023, the OECD formalised the OECD.AI Expert Group on AI Incidents to advance the development of i) a common AI incident reporting framework and ii) an AI Incidents Monitor (AIM).", "claim": "origin: the expert group formalised in January 2023 to develop the reporting framework and the monitor", "modality": "documented", "kind": "genealogy", "kernel": false, "const": false, "lineage": "maker: the OECD.AI expert group, McGovern support", "date": "2024-05 (OECD Artificial Intelligence Papers No. 16; approved by the DPC 2024-03-14)", "quote_ok": true, "checked_against": "saved text B5.pdf.txt (sha256 97db457717a855e2…), whitespace-normalised" }, "F51": { "source": "B5", "origin": "card", "locus": "§4 Draft definition of an AI incident", "quote": "An AI incident is an event, circumstance or series of events where the development, use or malfunction of one or more AI systems directly or indirectly leads to any of the following harms: (a) injury or harm to the health of a person or groups of people; (b) disruption of the management and operation of critical infrastructure; (c) violations of human rights or a breach of obligations under the applicable law intended to protect fundamental, labour and intellectual property rights; (d) harm to property, communities or the environment.", "claim": "the definition of an AI incident: development, use or malfunction leading directly or indirectly to harms (a)–(d)", "modality": "stipulation", "kind": "definition", "kernel": true, "const": true, "lineage": "definitions: incident and hazard", "date": "2024-05 (OECD Artificial Intelligence Papers No. 16; approved by the DPC 2024-03-14)", "quote_ok": true, "checked_against": "saved text B5.pdf.txt (sha256 97db457717a855e2…), whitespace-normalised" }, "F52": { "source": "B5", "origin": "card", "locus": "§5 Draft definition of an AI hazard", "quote": "An AI hazard is an event, circumstance or series of events where the development, use or malfunction of one or more AI systems could plausibly lead to an AI incident, i.e., any of the following harms:", "claim": "the definition of an AI hazard: an event that could plausibly lead to an AI incident", "modality": "stipulation", "kind": "definition", "kernel": true, "const": true, "lineage": "definitions: incident and hazard", "date": "2024-05 (OECD Artificial Intelligence Papers No. 16; approved by the DPC 2024-03-14)", "quote_ok": true, "checked_against": "saved text B5.pdf.txt (sha256 97db457717a855e2…), whitespace-normalised" }, "F53": { "source": "B5", "origin": "card", "locus": "§5 key clarifications", "quote": "Near misses are events that could have led to an AI incident and are therefore included under this definition of AI hazards.", "claim": "near misses are hazards", "modality": "stipulation", "kind": "definition", "kernel": false, "const": false, "lineage": "definitions: incident and hazard", "date": "2024-05 (OECD Artificial Intelligence Papers No. 16; approved by the DPC 2024-03-14)", "quote_ok": true, "checked_against": "saved text B5.pdf.txt (sha256 97db457717a855e2…), whitespace-normalised" }, "F54": { "source": "B5", "origin": "card", "locus": "§4 serious AI incident; AI disaster", "quote": "Serious AI incidents are a subset of AI incidents and are intended to include substantial, material incidents. … An AI disaster is a serious AI incident that disrupts the functioning of a community or a society and that may test or exceed its capacity to cope, using its own resources.", "claim": "the graded terms: serious AI incident (a subset) and AI disaster (a serious incident beyond a society's capacity to cope)", "modality": "stipulation", "kind": "definition", "kernel": false, "const": false, "lineage": "definitions: the graded terms", "date": "2024-05 (OECD Artificial Intelligence Papers No. 16; approved by the DPC 2024-03-14)", "quote_ok": true, "checked_against": "saved text B5.pdf.txt (sha256 97db457717a855e2…), whitespace-normalised (multi-part quote joined by ' … ', each part checked)" }, "F55": { "source": "B5", "origin": "card", "locus": "§5 serious AI hazard", "quote": "Serious AI hazards are a subset of AI hazards.", "claim": "serious AI hazards are a subset of hazards", "modality": "stipulation", "kind": "definition", "kernel": false, "const": false, "lineage": "definitions: the graded terms", "date": "2024-05 (OECD Artificial Intelligence Papers No. 16; approved by the DPC 2024-03-14)", "quote_ok": true, "checked_against": "saved text B5.pdf.txt (sha256 97db457717a855e2…), whitespace-normalised" }, "F56": { "source": "B5", "origin": "card", "locus": "§4 key clarifications", "quote": "Reputational harm to individuals and intangible harms such as hate speech and mis- and disinformation are included under (c) in relation to a breach of fundamental rights. … For example, violations of authors’ and personal rights fit into (c), while violations of industrial property rights are included under the concept of harm to property in (d).", "claim": "reputational harm, misinformation and violations of authors' rights fall under (c)", "modality": "stipulation", "kind": "definition", "kernel": false, "const": false, "lineage": "definitions: incident and hazard", "date": "2024-05 (OECD Artificial Intelligence Papers No. 16; approved by the DPC 2024-03-14)", "quote_ok": true, "checked_against": "saved text B5.pdf.txt (sha256 97db457717a855e2…), whitespace-normalised (multi-part quote joined by ' … ', each part checked)" }, "F57": { "source": "B5", "origin": "card", "locus": "Abstract", "quote": "These definitions aim to foster international interoperability while providing flexibility for jurisdictions to determine the scope of AI incidents and hazards they wish to address.", "claim": "the definitions' governing limit: interoperable, with each jurisdiction setting its own scope", "modality": "self-description", "kind": "limit", "kernel": true, "const": false, "lineage": "definitions: incident and hazard", "date": "2024-05 (OECD Artificial Intelligence Papers No. 16; approved by the DPC 2024-03-14)", "quote_ok": true, "checked_against": "saved text B5.pdf.txt (sha256 97db457717a855e2…), whitespace-normalised" }, "F58": { "source": "B5", "origin": "card", "locus": "§6 Conclusion ¶2", "quote": "Assessing the “seriousness” of an AI incident, harm, damage, or disruption (e.g., to determine whether an event is classified as an incident or a serious incident) is context-dependent and is also left for further discussion.", "claim": "seriousness is context-dependent and left open", "modality": "self-description", "kind": "limit", "kernel": false, "const": false, "lineage": "definitions: the graded terms", "date": "2024-05 (OECD Artificial Intelligence Papers No. 16; approved by the DPC 2024-03-14)", "quote_ok": true, "checked_against": "saved text B5.pdf.txt (sha256 97db457717a855e2…), whitespace-normalised" }, "F59": { "source": "B5", "origin": "card", "locus": "§3 ¶4", "quote": "This classification is based on a stocktaking exercise of over 30 frameworks, standards and legal instruments relevant to risk and harms", "claim": "the classification rests on a stocktaking of more than 30 frameworks, standards and legal instruments", "modality": "documented", "kind": "genealogy", "kernel": false, "const": false, "lineage": "definitions: incident and hazard", "date": "2024-05 (OECD Artificial Intelligence Papers No. 16; approved by the DPC 2024-03-14)", "quote_ok": true, "checked_against": "saved text B5.pdf.txt (sha256 97db457717a855e2…), whitespace-normalised" }, "F60": { "source": "B5", "origin": "card", "locus": "Acknowledgements ¶2", "quote": "Many experts provided invaluable feedback and suggestions, including Sean McGregor (Responsible AI Collaborative)", "claim": "the AI Incident Database's founder among those who gave feedback on the definitions", "modality": "documented", "kind": "attribution", "kernel": false, "const": false, "lineage": "neighbours: the AI Incident Database", "date": "2024-05 (OECD Artificial Intelligence Papers No. 16; approved by the DPC 2024-03-14)", "quote_ok": true, "checked_against": "saved text B5.pdf.txt (sha256 97db457717a855e2…), whitespace-normalised" }, "F61": { "source": "B6", "origin": "card", "locus": "oEmbed title (video, 29:48)", "quote": "Establishing Effective Serious Incident Reporting Frameworks – Jimmy Farrell | IASEAI 2025", "claim": "a conference talk on serious incident reporting frameworks (IASEAI 2025), admitted by title", "modality": "documented", "kind": "fact", "kernel": true, "const": false, "lineage": "reception: video", "date": "2025 (IASEAI 2025; title only)", "quote_ok": true, "checked_against": "saved text B6.txt (sha256 8c9f8a79bff55d6b…), whitespace-normalised" }, "F62": { "source": "B7", "origin": "own_page", "locus": "item title and summary", "quote": "AI Search Engines Fabricate News Citations, Undermining Publishers … A study by the Tow Center found that eight AI-powered search engines and chatbots, including ChatGPT, Perplexity and Elon Musk’s Grok3, frequently misquote or fabricate news citations, remixing publisher content without credit and driving down traffic. Over 60% of responses contained wrong or broken links, undermining journalistic credibility.[AI generated]", "claim": "the Tow Center's study of AI search citation, entered as an item; its summary machine-written", "modality": "documented", "kind": "fact", "kernel": true, "const": false, "lineage": "holding: the Tow Center study", "date": "2025-03-11 (event id 2025-03-11-8537; page as fetched 2026-10-29)", "quote_ok": true, "checked_against": "saved text B7.txt (sha256 4dc98555510507f2…), whitespace-normalised (multi-part quote joined by ' … ', each part checked)" }, "F63": { "source": "B7", "origin": "own_page", "locus": "label explanation; harm types", "quote": "Hence, the event meets the criteria for an AI Incident as it involves realized harm caused by AI system outputs.[AI generated] … Harm types Economic/PropertyReputationalPublic interest", "claim": "labelled an AI incident (realized harm from AI outputs), harm types economic/property, reputational, public interest", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "holding: the Tow Center study", "date": "2025-03-11 (event id 2025-03-11-8537; page as fetched 2026-10-29)", "quote_ok": true, "checked_against": "saved text B7.txt (sha256 4dc98555510507f2…), whitespace-normalised (multi-part quote joined by ' … ', each part checked)" }, "F64": { "source": "B7", "origin": "own_page", "locus": "articles list (25 articles, 2025-03-11 to 2025-03-17)", "quote": "Articles about this incident or hazard … New study finds that AI search tools are 60 percent inaccurate on average 2025-03-11 TechSpot", "claim": "the item is built from news articles about the study, the earliest dated 11 March 2025", "modality": "documented", "kind": "fact", "kernel": false, "const": false, "lineage": "holding: the Tow Center study", "date": "2025-03-11 (event id 2025-03-11-8537; page as fetched 2026-10-29)", "quote_ok": true, "checked_against": "saved text B7.txt (sha256 4dc98555510507f2…), whitespace-normalised (multi-part quote joined by ' … ', each part checked)" } }, "field_sources": { "B1": "OECD AI Policy Observatory — OECD AI Incidents Monitor, an evidence base for trustworthy AI", "B2": "OECD — AI risks and incidents", "B3": "OECD AI Policy Observatory — Overview and methodology of the AI Incidents and Hazards Monitor - OECD.AI", "B4": "OECD — Trends in AI incidents and hazards reported by the media | OECD", "B5": "OECD — Defining AI incidents and related terms (EN)", "B6": "YouTube·International Association for Safe & Ethical AI — Establishing Effective Serious Incident Reporting Frameworks – Jimmy Farrell", "B7": "OECD AI Policy Observatory — AI Search Engines Fabricate News Citations, Undermining Publishers (AIM item 2025-03-11-8537)", "W1": "Wikipedia — no article (OECD AI Incidents and Hazards Monitor, its variants, OECD.AI, OECD AI Policy Observatory)", "W2": "Wikipedia — OECD (no mention of the monitor)" }, "delta": { "T_vs_LB": "At 'oecd ai incidents and hazards monitor' (T1, six cards: B1–B6) the composition gives the monitor by its own pages and the OECD's papers. Composed from its cards: the maker and supporter (B3 F11, at a near grain: 'developed by the OECD.AI Policy Observatory in collaboration with' for 'initiated and is being developed by the OECD.AI expert group on AI incidents with the support of', CAP-1-05); real-time tracking and the evidence base (B3 F12; B4 F38; B1 F2); Event Registry and the 150,000 articles a day (B3 F13), with the AI-tagged events (F14, 'to detect AI-related events'); the definitions of incident and hazard (B5 F51–F53; B3 F19–F20), near misses as 'a near-harm' (F53). Assigned to another: the processing of 'over 150,000 news articles daily' is Event Registry's (F13: 'a news intelligence platform that monitors and aggregates world news by processing over 150 000 news articles daily'); T1-2 composes it as the monitor's ('It continuously aggregates and analyzes world news … processing over 150,000 news articles daily'). Substituted reading: the fourteen themes (T1-4, 'Thematic Classification: Media reports are categorized into 14 distinct risk themes') are the trends paper's analysis of the monitor's data (B4 F35), made by k-means clustering of the monitor's LLM-generated summaries into 25 clusters consolidated to 14 (F41, F42), each item assigned one theme (F43); T composes them as a classification the monitor performs; the methodology page gives no count (CAP-1-06, now read against B4); the paper's next steps include reorganising the record by theme (F46), not yet done. The theme names T lists beyond F35's span (cyberattacks and fraud, autonomous vehicles, LLM hallucinations, labour market disruptions) are in B4's executive summary outside the ledgered spans. Limit dropped: T1-2's 'Real-Time Tracking … continuously' against the daily pipeline on events one to four days old (F31); T1-5 drops harm (d), 'harm to property, communities or the environment', and 'directly or indirectly' and 'series of events' from the incident (F51); T1-6 drops 'malfunction' from the hazard (F52). Force raised: T1-3, 'empirical data on how AI risks materialize globally', against the field's own limits: 'only represent a subset of all AI incidents and hazards worldwide … a useful starting point' (F15), 'reflect media and public interest rather than comprehensive or systematic tracking' (F36), errors of inclusion (F37), trends that do not measure severity (F44), and information the OECD 'does not independently verify' (F24). Composed without attribution in the body, at the source's grain: the definitions (T1-5, T1-6) carry markers but not the paper (F48, F18), which is card 5. Available, not composed: the self-description 'Automated media discourse monitor of AI incidents and hazards (Beta)' (F1); the LLM classification since November 2024 (F25–F27), the o3-mini metadata from the top three articles (F28–F29), grouping by summary similarity (F30) and the earlier 'traditional machine learning techniques' (F34), which the Capture Registry names as the classifier left out (CAP-1-02, CAP-1-07); the '[AI generated]' marks and the further classes on the monitor's own page (F8, F9); the unit as an event reported by one or more articles, on Microsoft Azure (F6); the count (F7) and the two daily rates (F32, F33); the trend, more attention and a falling share (F4, F40); the top 50% of outlets by Alexa rank (F39); the disclaimers (F5, F17, F23, F24); the graded terms (F54, F55, F58), the place of reputational harm, misinformation and authors' rights under (c) (F56), the jurisdictional limit (F57), the AI system definition (F21), the paper's authors, approval and stocktaking (F49, F59) and the expert group's formalisation in January 2023 (F50); the former name (F22); next steps (F16, F46); the common reporting framework and the scarcity of public sources (F45); the AI Incident Database's founder among the definitions' reviewers (F60); the OECD topic page's snippet (F10, card 2, unreachable); the video (F61, title only). Content replaced by use: the access line (T1-7) and the closer (T1-8), which offers the common reporting framework guidelines (F45, F16) as a next step. Not surfaced by T: the holding B7 (the Tow Center item, F62–F64), an own page the MMRS record cites, admitted beside the cards by the operator's ruling. Wikipedia has no article (W1) and the OECD article does not mention the monitor (W2).", "T_vs_LB_reading": "v0.9 §0.5: the first diff is already ontological. The field's ontology holds the monitor as an automated media monitor whose classifier and record are machine-made (F1, F26, F28, F8) and whose evidence is bounded by what the media report (F15, F36); T's holds it as a tracking tool that aggregates and analyses the news and categorises it into themes. What T keeps of its cards is the monitor's inputs and definitions; what it drops is the machine in the method and the limits on the evidence, and it composes a paper's analysis of the monitor's machine-written summaries as the monitor's own classification. The row records this as observed and claims nothing about the composer's internals (§0.5).", "LB_vs_LBA": "Admission adds, in one claim space with the field's: the monitor as an automated monitor of AI harm events in the news, run by OECD.AI (MMRS-01, MMRS-19), and as itself a machine reception of the news, its summaries compositions and the articles their sources (MMRS-12), the record not preserving the original outputs (MMRS-25) — the field states the machine-written record (F28, F8) and the archive reads its consequence; the composition layer's reading against it, 'as an aggregator of the news' (CAP-1-03); what its selection leaves out, 'A harm no outlet reports has no event to cluster' (MMRS-11), with admission requiring harm (MMRS-24) and entering by reputable outlets (MMRS-10), which extends the field's own limits (F15, F36) to the unit; the record kept apart from the entity, with its unchecked launch date and licence (MMRS-31, MMRS-26, MMRS-27); the neighbours, the automated counterpart of the editor-resolved AI Incident Database and one of the census's three harm repositories (MMRS-13, MMRS-15, MMRS-16, MMRS-29); the holding, a research study entering a harm monitor as an event (MMRS-14, MMRS-28), read against its own page (B7, F62–F64); the archive's class-level distinction of media monitors, which 'track coverage of events', from its own instrument of the retrieval layer (#637, O637-01); and the composition layer's observation at the entity's address (CAP-1-01–08, MMRS-30). Coherence (P_coh, composed each with its source): the MMRS record's operator OECD.AI (MMRS-18) beside the methodology page's expert group with McGovern support (F11), not in contradiction; the count about 18,579 observed 2026-10-28 (MMRS-22) and the same count on the page 2026-10-29 (F7), beside the page's two daily rates (F32, F33; internally_inconsistent in B3). Restated lineages (§3.6, the own page keeping the earlier date): MMRS-03, MMRS-19 with F1; MMRS-04, MMRS-05 with F51, F52, F18; MMRS-06 with F14; MMRS-02, MMRS-07, MMRS-21 with F25, F26, F28; MMRS-08 with F30; MMRS-09, MMRS-23 with F31; MMRS-17 with F22; MMRS-20 with F6; MMRS-22 with F7; MMRS-25 with F28; MMRS-14, MMRS-28 with F62. The measure is in row.measure.", "alignment": { "T1_cards_claims": [ "F1", "F2", "F3", "F4", "F5", "F6", "F7", "F8", "F9", "F10", "F11", "F12", "F13", "F14", "F15", "F16", "F17", "F18", "F19", "F20", "F21", "F22", "F23", "F24", "F25", "F26", "F27", "F28", "F29", "F30", "F31", "F32", "F33", "F34", "F35", "F36", "F37", "F38", "F39", "F40", "F41", "F42", "F43", "F44", "F45", "F46", "F47", "F48", "F49", "F50", "F51", "F52", "F53", "F54", "F55", "F56", "F57", "F58", "F59", "F60", "F61" ], "T1_composed": [ "F2", "F11", "F12", "F13", "F14", "F19", "F20", "F35", "F38", "F51", "F52", "F53" ], "rule": "a field claim counts as composed when T states it at its source's grain or coarser; the typology of what is not is in T_vs_LB (§6.2). T1_cards_claims are the claims of B1–B6 (B7 is an own page T did not surface)" } }, "kernel": [ { "K": "K1", "claim": "`MMRS-01`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "stipulation", "sense": "what it is: an automated media monitor", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F1, F38)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K2", "claim": "`MMRS-19`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "stipulation", "sense": "what it is: an automated media monitor", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F1)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K3", "claim": "`MMRS-03`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "attributed", "sense": "what it is: an automated media monitor", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F1)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K4", "claim": "`MMRS-17`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "name and terms", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F22)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K5", "claim": "`MMRS-18`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "maker: the OECD.AI expert group, McGovern support", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "qualified claim (F11: the expert group, McGovern support)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K6", "claim": "`MMRS-05`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "definitions: incident and hazard", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F18)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K7", "claim": "`MMRS-04`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "attributed", "sense": "definitions: incident and hazard", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F51, F52)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K8", "claim": "`MMRS-20`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "the unit: the event and its articles", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F6, F51, F52)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K9", "claim": "`MMRS-06`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "intake: Event Registry's news events", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F14)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K10", "claim": "`MMRS-10`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "interpretation", "sense": "what its selection leaves out", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "qualified claim (F13: reputable international news outlets)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K11", "claim": "`MMRS-02`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "method: LLM classification since November 2024", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F25, F26, F28)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K12", "claim": "`MMRS-07`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "method: LLM classification since November 2024", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F26, F28)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K13", "claim": "`MMRS-08`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "method: LLM classification since November 2024", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F30)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K14", "claim": "`MMRS-21`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "method: LLM classification since November 2024", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F25, F26, F28)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K15", "claim": "`MMRS-09`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "cadence: daily, one to four days", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F31)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K16", "claim": "`MMRS-23`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "cadence: daily, one to four days", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F31)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K17", "claim": "`MMRS-25`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "the record: machine-written", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "qualified claim (F28)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K18", "claim": "`MMRS-12`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "interpretation", "sense": "the monitor as machine reception", "qualifiers carried": "MMRS-11, MMRS-24", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "qualified claim (F38: tracks media-reported incidents; F28: the metadata LLM-generated)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K19", "claim": "`CAP-1-03`", "source": "Capture Registry, 2026-10-29", "M_src": "interpretation", "sense": "the monitor as machine reception", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "qualified claim (F38)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K20", "claim": "`MMRS-22`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "count", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (F7: the same count a day later)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K21", "claim": "`MMRS-24`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "stipulation", "sense": "what its selection leaves out", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates F51, F52 as an admission rule)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K22", "claim": "`MMRS-11`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "interpretation", "sense": "what its selection leaves out", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "qualified claim (F15, F36: a subset; what journalists know)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K23", "claim": "`MMRS-31`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "stipulation", "sense": "the record kept apart from the entity", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing distinction (the record kept apart from the entity)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K24", "claim": "`MMRS-27`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "the record kept apart from the entity", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing distinction (lifecycle apart from observation)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K25", "claim": "`MMRS-26`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "self-description", "sense": "the record kept apart from the entity", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing distinction (unchecked: launch date, licence; cf. F50, F34, F47)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K26", "claim": "`CAP-1-06`", "source": "Capture Registry, 2026-10-29", "M_src": "self-description", "sense": "themes: the 2026 paper's analysis", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing finding (an observation of the composition at the entity's address)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K27", "claim": "`MMRS-13`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "interpretation", "sense": "neighbours: the AI Incident Database", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing distinction (the field names no counterpart)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K28", "claim": "`MMRS-16`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "interpretation", "sense": "neighbours: the AI Incident Database", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing distinction", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K29", "claim": "`MMRS-29`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "neighbours: the AI Incident Database", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing distinction", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K30", "claim": "`MMRS-15`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "interpretation", "sense": "neighbours: the AI Incident Database", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing distinction", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K31", "claim": "`MMRS-14`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "holding: the Tow Center study", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates B7, F62)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K32", "claim": "`MMRS-28`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "holding: the Tow Center study", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates B7, F62)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K33", "claim": "`O637-01`", "source": "#637", "M_src": "interpretation", "sense": "relation: media monitors and the archive's instruments", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "qualified claim (F1: a media discourse monitor)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K34", "claim": "`CAP-1-01`", "source": "Capture Registry, 2026-10-29", "M_src": "documented", "sense": "reception: the composition layer", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing finding (an observation of the composition at the entity's address)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K35", "claim": "`CAP-1-05`", "source": "Capture Registry, 2026-10-29", "M_src": "interpretation", "sense": "reception: the composition layer", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing finding (an observation of the composition at the entity's address)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K36", "claim": "`CAP-1-02`", "source": "Capture Registry, 2026-10-29", "M_src": "documented", "sense": "reception: the composition layer", "qualifiers carried": "CAP-1-06", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing finding (an observation of the composition at the entity's address)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K37", "claim": "`CAP-1-07`", "source": "Capture Registry, 2026-10-29", "M_src": "interpretation", "sense": "reception: the composition layer", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing finding (an observation of the composition at the entity's address)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K38", "claim": "`CAP-1-04`", "source": "Capture Registry, 2026-10-29", "M_src": "documented", "sense": "reception: the composition layer", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing finding (an observation of the composition at the entity's address)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K39", "claim": "`CAP-1-08`", "source": "Capture Registry, 2026-10-29", "M_src": "interpretation", "sense": "reception: the composition layer", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "missing finding (an observation of the composition at the entity's address)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" }, { "K": "K40", "claim": "`MMRS-30`", "source": "MMRS record (2026-10-28, corrected 2026-10-29)", "M_src": "documented", "sense": "reception: the composition layer", "qualifiers carried": "—", "f (source's own falsifiers)": "— (none stated; watch: a later composition at the address that states or drops the claim, attributed or not, uptake §8.3)", "contrast": "none (restates the capture record)", "resolver": "a later capture at 'oecd ai incidents and hazards monitor' (ADDR-de46fae4a143) on Google AI Overview, seated in the Capture Registry (§8.3)" } ], "kernel_T": [ { "claim": "F1", "note": "'Automated media discourse monitor … (Beta)': the monitor's self-description (absent)" }, { "claim": "F26", "note": "LLM classification since November 2024, GPT-4o mini and GPT-4o (absent; CAP-1-02)" }, { "claim": "F28", "note": "the record's metadata LLM-generated by o3-mini from the top three articles (absent; CAP-1-02)" }, { "claim": "F8", "note": "each summary and label explanation marked '[AI generated]' on the monitor's page (absent)" }, { "claim": "F15", "note": "a subset of all incidents and hazards (limit dropped against T1-3)" }, { "claim": "F36", "note": "media and public interest, what journalists know and deem newsworthy (limit dropped against T1-3)" }, { "claim": "F24", "note": "the OECD does not independently verify the third-party information (absent)" }, { "claim": "F31", "note": "daily, on events one to four days old (limit dropped against T1-2's 'Real-Time … continuously')" }, { "claim": "F41", "note": "the fourteen themes are the trends paper's clusters of LLM-generated summaries (substituted as the monitor's classification, T1-4)" }, { "claim": "F39", "note": "the top 50% of outlets by Alexa traffic rank (absent)" }, { "claim": "F54", "note": "the graded terms: serious incident, AI disaster, serious hazard (absent)" }, { "claim": "F7", "note": "the count, about 18,579 (absent)" }, { "claim": "F48", "note": "the definitions' source, the OECD paper of May 2024, not named in the body (its card is card 5)" } ], "ledger": { "archive": "datasets/negative-of-the-negative/v2/ledgers/oecd-ai-incidents-monitor/ledger-archive.json", "field": "datasets/negative-of-the-negative/v2/ledgers/oecd-ai-incidents-monitor/field.json", "selection": "datasets/negative-of-the-negative/v2/traversal/oecd-ai-incidents-monitor/", "config": "datasets/negative-of-the-negative/v2/panel/configs/oecd-ai-incidents-monitor.json", "intake": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/", "audit": null }, "ontology": { "composition": "equal", "composition_note": "one claim space; the archive's readings (the monitor as a machine reception of the news; what its selection leaves out) composed as senses of the entity beside the field's statements they draw on", "case": { "value": "equal", "statement": "The MMRS record holds the field's entity as the entity: it describes the monitor from its own pages (the methodology page and the monitor's page), marking each field's source and basis, and composes nothing that denies a field presupposition. What it adds lies on the field's object: the record's machine-written wording read as the monitor being itself a machine reception of the news, its summaries compositions and the articles their sources; what its selection leaves out (a harm no outlet reports has no event); the record kept apart from the entity; the neighbours in the harm basin; the holding; and the Capture Registry's observation that the composition layer left the classifier out. The ontology the archive uses is available in the field itself: the methodology page states that the metadata, title and summary are 'LLM-generated' (F28), the monitor's page marks them '[AI generated]' (F8), the trends paper clusters the monitor's 'LLM-generated summaries' (F41) and bounds its evidence by what journalists know (F36). The archive's reading draws the consequence of the field's own statements; the law of the suppressed witness does not apply (§0.10: not where the field holds the ontology). Not mixed.", "loci": [ { "dep": "MMRS", "locus": "article, kernel s1 (articles.py)", "quote": "The OECD AI Incidents and Hazards Monitor (AIM) is an automated monitor of AI harm events in the news, run by OECD.AI." }, { "dep": "MMRS", "locus": "article, What its selection leaves out s3", "quote": "The record's own wording is machine-written from the articles, so the monitor is itself a machine reception of the news: its summaries are compositions, and the articles are their sources." }, { "dep": "MMRS", "locus": "census.json, schema_note", "quote": "each page has two nodes, the MMRS record and the entity it describes; relations are typed; each assertion carries a source and a basis (operator statement, third-party report, our observation, our reading); lifecycle is kept apart from the observation that saw it" }, { "dep": "capture", "locus": "oecd-ai-incidents-and-hazards-monitor-aio-20261029, analysis s1", "quote": "The monitor is composed from its own methodology page with its inputs and definitions right and its classifier left out, the part of its method that makes each record a machine composition." }, { "dep": "B3", "locus": "Methodology step 5 (field)", "quote": "This metadata is LLM-generated (OpenAI’s o3-mini) from the top three articles of each event, selected from different news outlets." }, { "dep": "B4", "locus": "Annex A (field)", "quote": "LLM-generated summaries of AI incidents and hazards were collected and transformed into a semantic representation capturing the underlying concepts." } ] }, "E_F": { "type": "C (public entity: a monitor)", "definition": "an automated media discourse monitor of AI incidents and hazards (beta), initiated and developed by the OECD.AI expert group on AI incidents with the support of the Patrick J. McGovern Foundation, previously the AI Incidents Monitor, whose unit is the event (an incident or hazard in the OECD's terms of May 2024, reported by one or more news articles), drawn from Event Registry's AI-tagged news events from reputable outlets, classified since November 2024 by LLMs (GPT-4o mini, GPT-4o) and recorded in LLM-generated titles, summaries and metadata (o3-mini), daily; its evidence a subset bounded by what the media report (B1–B7)", "claims": [ "F1", "F11", "F22", "F6", "F51", "F52", "F13", "F14", "F26", "F28", "F31", "F15", "F36" ] }, "E_A": { "definition": "the same monitor, in the MMRS meta-aggregator's record: an automated monitor of AI harm events in the news, run by OECD.AI, whose records are machine-written from the articles, so that the monitor is itself a machine reception of the news, its summaries compositions and the articles their sources; admitting an event when reputable outlets report it, so that a harm no outlet reports has no event; the automated counterpart of the editor-resolved AI Incident Database in the harm basin; a record kept apart from the entity; observed by the Capture Registry at its name, where the composition layer gave its inputs and definitions and left its classifier out; and a member of the class of media monitors, which the archive sets apart from its own instrument of the retrieval layer (#637)", "claims": [ "MMRS-01", "MMRS-12", "MMRS-25", "MMRS-10", "MMRS-11", "MMRS-13", "MMRS-31", "CAP-1-07", "CAP-1-02", "O637-01" ], "defining_papers": [ 637 ], "defining_records": [ "MMRS record (census.json rows[slug == 'oecd-ai-incidents-monitor'] and schema_note, articles.py ARTICLES['oecd-ai-incidents-monitor'], with the sentences of ARTICLES['ai-incident-database'] and ARTICLES['aiaaic-repository'] that name it; rendered aggregators/oecd-ai-incidents-monitor/index.html)", "Capture Registry oecd-ai-incidents-and-hazards-monitor-aio-20261029 (OBS-23f50c1b7e04)" ] }, "rho": { "value": "extends", "status": "read from the MMRS record and the capture record (§1.2a); no deposit defines the archive's entity; the one deposit admitted (#637) bears on the class (media monitors), not on the monitor by name", "statement": "the archive reads E_F as the entity and adds senses and limits on the field's own object: the monitor as a machine reception of the news (the consequence of the machine-written record the field states), what its selection leaves out (no event for an unreported harm), the record kept apart from the entity and its unchecked items, the neighbours and the holding, the class-level distinction of media monitors from the archive's instrument, and the composition layer's observed treatment of the entity", "claims": [ "MMRS-12", "MMRS-25", "MMRS-11", "MMRS-10", "MMRS-31", "MMRS-26", "MMRS-13", "MMRS-14", "O637-01", "CAP-1-03", "CAP-1-07" ], "guard": "ρ = extends claims an addition and no more: MMRS-10, MMRS-11 and MMRS-12 are the record's readings (interpretation), not findings that the monitor is defective; O637-01 is a distinction drawn at the class, not a claim on the monitor by name; where an archive claim meets a field claim (MMRS-18 beside F11; MMRS-22 beside F7, F32, F33) both are composed under P_coh in the one claim space", "considered": [ { "value": "constructs", "set_aside": "the archive does not read the monitor as made by its receiving texts" }, { "value": "revises", "set_aside": "the entity is not the archive's own" }, { "value": "witness (suppressed witness)", "set_aside": "the field holds the ontology (§0.10): the machine-written record is the field's own statement (F8, F28, F41)" } ] }, "translation": [ { "lineage": "what it is: an automated media monitor", "field_claims": [ "F1", "F38" ], "received": "F1 (B1, header (subtitle under 'AIM: AI Incidents and Hazards Monitor')): \"Automated media discourse monitor of AI incidents and hazards (Beta)\"; F38 (B4, Executive summary ¶7): \"The OECD.AI Incidents and Hazards Monitor (AIM) tracks these media-reported AI incidents and hazards in real-time and leverages the OECD common reporting framework to categorise them consistently.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-01, MMRS-19, MMRS-03", "archive_claims": [ "MMRS-01", "MMRS-19", "MMRS-03" ] }, { "lineage": "name and terms", "field_claims": [ "F22", "F47" ], "received": "F22 (B3, Information Transparency Disclosures: Background): \"Your use of the OECD AI incidents and hazards monitor (previously AI Incidents Monitor) (“AIM”) is subject to the terms and conditions found at www.oecd.org/termsandconditions.\"; F47 (B4, rights page): \"Attribution 4.0 International (CC BY 4.0)\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-17", "archive_claims": [ "MMRS-17" ] }, { "lineage": "maker: the OECD.AI expert group, McGovern support", "field_claims": [ "F11", "F50" ], "received": "F11 (B3, Overview ¶1 s1): \"The AI Incidents and Hazards Monitor (AIM) was initiated and is being developed by the OECD.AI expert group on AI incidents with the support of the Patrick J. McGovern Foundation.\"; F50 (B5, Executive summary ¶1): \"In January 2023, the OECD formalised the OECD.AI Expert Group on AI Incidents to advance the development of i) a common AI incident reporting framework and ii) an AI Incidents Monitor (AIM).\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-18", "archive_claims": [ "MMRS-18" ] }, { "lineage": "purpose: an evidence base for policy", "field_claims": [ "F2", "F3", "F12", "F10" ], "received": "F2 (B1, introduction ¶1 s2): \"The OECD AI Incidents and Hazards Monitor (AIM) documents AI incidents and hazards to help policymakers, AI practitioners, and all stakeholders worldwide gain valuable insights into the risks and harms of AI systems.\"; F3 (B1, introduction ¶1 s3): \"Over time, AIM will help to show risk patterns and establish a collective understanding of AI incidents and hazards and their multifaceted nature, serving as an important tool for trustworthy AI.\"; F12 (B3, Overview ¶1 s2): \"The goal of the AIM is to track actual AI incidents and hazards, as defined by the OECD, in real time and provide the evidence-base to inform related AI policy discussions.\"; F10 (B2, card snippet (page HTTP 403)): \"All actors must monitor AI incidents and hazards to mitigate risks As AI adoption continues to grow, successful risk mitigation will require a solid evidence ba...\"", "receiving_text": "—", "translation": "untranslated: composed as received (no archive claim in this lineage)", "archive_claims": [] }, { "lineage": "definitions: incident and hazard", "field_claims": [ "F48", "F49", "F59", "F18", "F51", "F19", "F52", "F20", "F53", "F56", "F21", "F57" ], "received": "F48 (B5, cover; Foreword ¶1): \"May 2024 No. 16 … This report provides a definition of AI incident and related terminology. These definitions aim to clarify what constitutes an AI incident, an AI hazard, and associated terminology without being overly prescriptive. This document will inform the development of a common AI incidents reporting framework and its application through the OECD AI Incidents Monitor (AIM).\"; F49 (B5, Foreword ¶3–4): \"The report was written by Karine Perset and Luis Aranda under the supervision of Audrey Plonk, Deputy Director of the OECD Science Technology and Innovation Directorate. … This paper was approved and declassified by written procedure by the OECD Digital Policy Committee (DPC) on 14 March 2024 and prepared for publication by the OECD Secretariat.\"; F59 (B5, §3 ¶4): \"This classification is based on a stocktaking exercise of over 30 frameworks, standards and legal instruments relevant to risk and harms\"; F18 (B3, Definitions ¶1): \"Published in May 2024, the paper Defining AI incidents and related terms, defines an event where the development or use of an AI system results in actual harm as an “AI incident”, while an event where the development or use of an AI system is potentially harmful is termed an “AI hazard”.\"; F51 (B5, §4 Draft definition of an AI incident): \"An AI incident is an event, circumstance or series of events where the development, use or malfunction of one or more AI systems directly or indirectly leads to any of the following harms: (a) injury or harm to the health of a person or groups of people; (b) disruption of the management and operation of critical infrastructure; (c) violations of human rights or a breach of obligations under the applicable law intended to protect fundamental, labour and intellectual property rights; (d) harm to property, communities or the environment.\"; F19 (B3, Definitions: AI incident): \"An AI incident is an event, circumstance or series of events where the development, use or malfunction of one or more AI systems directly or indirectly leads to any of the following harms:\"; F52 (B5, §5 Draft definition of an AI hazard): \"An AI hazard is an event, circumstance or series of events where the development, use or malfunction of one or more AI systems could plausibly lead to an AI incident, i.e., any of the following harms:\"; F20 (B3, Definitions: AI hazard): \"An AI hazard is an event, circumstance or series of events where the development, use or malfunction of one or more AI systems could plausibly lead to an AI incident, i.e., any of the following harms:\"; F53 (B5, §5 key clarifications): \"Near misses are events that could have led to an AI incident and are therefore included under this definition of AI hazards.\"; F56 (B5, §4 key clarifications): \"Reputational harm to individuals and intangible harms such as hate speech and mis- and disinformation are included under (c) in relation to a breach of fundamental rights. … For example, violations of authors’ and personal rights fit into (c), while violations of industrial property rights are included under the concept of harm to property in (d).\"; F21 (B3, Definitions: AI system): \"These definitions were built based on the definition of an AI system as described in the OECD recommendation on AI of 2019 and revised in 2024. An AI system is a machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments.\"; F57 (B5, Abstract): \"These definitions aim to foster international interoperability while providing flexibility for jurisdictions to determine the scope of AI incidents and hazards they wish to address.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-05, MMRS-04", "archive_claims": [ "MMRS-05", "MMRS-04" ] }, { "lineage": "definitions: the graded terms", "field_claims": [ "F54", "F55", "F58" ], "received": "F54 (B5, §4 serious AI incident; AI disaster): \"Serious AI incidents are a subset of AI incidents and are intended to include substantial, material incidents. … An AI disaster is a serious AI incident that disrupts the functioning of a community or a society and that may test or exceed its capacity to cope, using its own resources.\"; F55 (B5, §5 serious AI hazard): \"Serious AI hazards are a subset of AI hazards.\"; F58 (B5, §6 Conclusion ¶2): \"Assessing the “seriousness” of an AI incident, harm, damage, or disruption (e.g., to determine whether an event is classified as an incident or a serious incident) is context-dependent and is also left for further discussion.\"", "receiving_text": "—", "translation": "untranslated: composed as received (no archive claim in this lineage)", "archive_claims": [] }, { "lineage": "the unit: the event and its articles", "field_claims": [ "F6" ], "received": "F6 (B1, chart block, note): \"Note: An AI incident or hazard can be reported by one or more news articles covering the same event. Data processing powered by Microsoft Azure using data from Event Registry.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-20", "archive_claims": [ "MMRS-20" ] }, { "lineage": "intake: Event Registry's news events", "field_claims": [ "F13", "F14", "F39" ], "received": "F13 (B3, Overview ¶2 s1–2): \"The AIM detects AI incidents and hazards from reputable international news outlets. The data is provided by Event Registry (ER), a news intelligence platform that monitors and aggregates world news by processing over 150 000 news articles daily.\"; F14 (B3, Overview ¶2 s3–5): \"Event Registry detects events, which are clusters of news articles reporting on the same happening. These events are tagged with related concepts. The events provided by Event Registry to AIM are limited to those that include AI concepts.\"; F39 (B4, §1 Introduction ¶6): \"Currently the data rely on reports from the top 50% of media outlets, based on the Alexa traffic rankings (OECD.AI, 2025[6]). These reports are aggregated by Event Registry, a news intelligence platform which monitors over 150 000 news articles daily.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-06", "archive_claims": [ "MMRS-06" ] }, { "lineage": "method: LLM classification since November 2024", "field_claims": [ "F25", "F26", "F27", "F34", "F30" ], "received": "F25 (B3, Methodology ¶1–2): \"Since November 2024, incidents and hazards are identified and added to the AIM through the following steps. Historical incidents and hazards, collected prior to November 2024, are currently being retrospectively annotated using the same methodology to ensure consistency.\"; F26 (B3, Methodology step 2): \"LLMs are used to classify AI-related events into three categories: AI incidents, AI hazards or unrelated events, as defined above. This classification is based on the event summary provided from Event Registry. The classification follows a two-step process: first, a smaller model (OpenAI’s GPT-4o mini) filters out events that seem unrelated to AI incidents and hazards; next, a larger model (OpenAI’s GPT-4o) reclassifies the remaining events to confirm whether they are incidents or hazards.\"; F27 (B3, Methodology step 3): \"Individual news articles are classified into incidents and hazards using LLMs, following the same process as for events. In this case, however, article content is used instead of event summaries to enhance reliability. … If some articles are classified differently, the event will adopt the most common classification among them.\"; F34 (B3, Methodology, last ¶): \"Prior to November 2024, AI incidents and hazards were classified and included on the AIM using traditional machine learning techniques.\"; F30 (B3, Methodology step 6): \"Similar events are grouped by analysing their summaries and identifying related events in AIM using cosine similarity. For each new incident or hazard, the top 10 closest summaries (i.e., “embeddings”) from existing events in AIM are identified. An LLM (OpenAI’s o3-mini) is then used to assess whether any of these similar events can be clustered together.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-02, MMRS-07, MMRS-08, MMRS-21", "archive_claims": [ "MMRS-02", "MMRS-07", "MMRS-08", "MMRS-21" ] }, { "lineage": "cadence: daily, one to four days", "field_claims": [ "F31" ], "received": "F31 (B3, Update frequency ¶1–2): \"The pipeline runs daily, processing events from one to four days ago. … Articles appearing after this timeframe are likely to be misclassified as a separate event.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-09, MMRS-23", "archive_claims": [ "MMRS-09", "MMRS-23" ] }, { "lineage": "the record: machine-written", "field_claims": [ "F28", "F29", "F8", "F9" ], "received": "F28 (B3, Methodology step 5): \"Each event is enriched with relevant metadata, including an event title, summary, the harm type, severity, affected stakeholders, and the country where it occurred, among others. This metadata is LLM-generated (OpenAI’s o3-mini) from the top three articles of each event, selected from different news outlets.\"; F29 (B3, Methodology step 5, continued): \"Articles are ranked based on their outlets’ ranking in their respective country. This approach ensures a balance between the popularity of the outlets and fair geographical representation. Currently, the summary and metadata are generated at the time the incident or hazard is added to the AIM.\"; F8 (B1, result list, the first item's explanation (as fetched 2026-10-29)): \"Why's our monitor labelling this an incident or hazard? … Therefore, the event is best classified as an AI Hazard due to the credible risk of AI exacerbating economic disparities and job losses in the future.[AI generated]\"; F9 (B1, result list, an item's explanation (as fetched 2026-10-29)): \"It is not Complementary Information because the article is not about responses or updates to a past incident, nor is it Beneficial Use or Unrelated.[AI generated]\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-25", "archive_claims": [ "MMRS-25" ] }, { "lineage": "count", "field_claims": [ "F7", "F32", "F33" ], "received": "F7 (B1, results line): \"Results: About 18579 incidents & hazards\"; F32 (B3, Summary table, Identification and classification): \"On average, about 1 000 AI events are analysed daily, resulting in around 10 AI incidents or hazards.\"; F33 (B3, Summary table, note): \"the AIM detects on average 30 incidents and hazards per day.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-22", "archive_claims": [ "MMRS-22" ] }, { "lineage": "trend: counts and shares", "field_claims": [ "F4", "F40" ], "received": "F4 (B1, introduction ¶1 s4): \"AI incidents seem to be getting more media attention lately, but they've actually gone down as a share of all AI news (see chart below!).\"; F40 (B4, §1 Introduction ¶8): \"Between 2022 and 2025, the number of AI incidents and hazards reported by the media rose steadily, from 92 to 324 per month on average (Figure 1.1, Panel A). However, their share relative to total AI news declined, from 3.2% in 2022 to around 2.5% in 2025 (Figure 1.1, Panel B).\"", "receiving_text": "—", "translation": "untranslated: composed as received (no archive claim in this lineage)", "archive_claims": [] }, { "lineage": "limits: a subset of what happens", "field_claims": [ "F15", "F36", "F37", "F44" ], "received": "F15 (B3, Overview ¶3): \"While recognising the likelihood that these incidents and hazards only represent a subset of all AI incidents and hazards worldwide, these publicly reported incidents and hazards nonetheless provide a useful starting point for building the evidence base.\"; F36 (B4, Executive summary ¶5): \"The findings reflect media and public interest rather than comprehensive or systematic tracking of AI incidents and hazards. Importantly, media reports capture only what journalists are aware of and deemed newsworthy.\"; F37 (B4, Executive summary ¶6): \"Conversely, reliance on media sources may introduce errors of inclusion, as some reported incidents may be mischaracterised, exaggerated, or inaccurately attributed to AI due to limited background or technical understanding.\"; F44 (B4, §2, after Figure 2.1): \"These trends do not measure how severe incidents are, the types of harm involved, or the number of affected stakeholders, and should not be read as indicators of these factors.\"", "receiving_text": "—", "translation": "untranslated: composed as received (no archive claim in this lineage)", "archive_claims": [] }, { "lineage": "limits: the monitor's own disclaimers", "field_claims": [ "F5", "F23", "F24", "F17" ], "received": "F5 (B1, introduction ¶2): \"The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.\"; F23 (B3, Information Transparency Disclosures: Third-Party Information): \"the AIM is populated with news articles from various third-party outlets and news aggregators with which the OECD has no affiliation.\"; F24 (B3, Information Transparency Disclosures: Errors and Omissions): \"The OECD cannot guarantee and does not independently verify the accuracy, completeness, or validity of third-party information provided in the AIM.\"; F17 (B3, Overview ¶5): \"The data collection and analysis for the AIM is done to ensure, to the best extent possible, the reliability, objectivity and quality of the information for AI incidents and hazards.\"", "receiving_text": "—", "translation": "untranslated: composed as received (no archive claim in this lineage)", "archive_claims": [] }, { "lineage": "themes: the 2026 paper's analysis", "field_claims": [ "F35", "F41", "F42", "F43" ], "received": "F35 (B4, Abstract): \"This paper analyses live data from the OECD.AI Incidents and Hazards Monitor (AIM), classifying media reports into 14 themes, including synthetic media, child safety, cyberattacks, privacy, and health. It identifies three distinct patterns of media coverage – increasing, intermittent, and declining – and traces how media attention shifts across themes over time.\"; F41 (B4, Annex A, Identifying key thematic clusters ¶1): \"LLM-generated summaries of AI incidents and hazards were collected and transformed into a semantic representation capturing the underlying concepts. Using a k-means clustering algorithm, these reports were classified into 25 thematic clusters each capturing semantically similar risk characteristics. … This resulted in 14 distinct themes.\"; F42 (B4, Annex A, Clustering ¶2): \"Since these methods gave inconsistent results, the authors chose 25 clusters based on a qualitative keywords review.\"; F43 (B4, §2 ¶2): \"To maintain analytical clarity and avoid double counting, each media-reported incident or hazard is assigned to the single most relevant theme based on the available information.\"", "receiving_text": "—", "translation": "untranslated: composed as received beside CAP-1-06", "archive_claims": [ "CAP-1-06" ] }, { "lineage": "place: the common reporting framework and proxies", "field_claims": [ "F45" ], "received": "F45 (B4, Executive summary ¶2–4): \"In early 2025, the members of the Global Partnership on AI (GPAI) approved a common framework for reporting AI incidents, designed as an international data-sharing template for public and private organisations across jurisdictions and sectors … At present, there are few public and comparable sources of information on AI incidents and hazards. … Proxy indicators – such as court rulings on AI-related cases or media reports – can offer partial insights into where and what types of incidents are occurring.\"", "receiving_text": "—", "translation": "untranslated: composed as received (no archive claim in this lineage)", "archive_claims": [] }, { "lineage": "next steps: open submission, other sources", "field_claims": [ "F16", "F46" ], "received": "F16 (B3, Overview ¶4): \"An open submission process following the common reporting framework for AI incidents is currently being developed to complement AI incidents and hazards from news articles. Other next steps include complementing incidents and hazards with court rulings and decisions by supervisory authorities wherever they exist.\"; F46 (B4, §6 Conclusion ¶4): \"Next steps could include allowing open submissions to complement media-based monitoring and reorganising incidents and hazards visually by theme to support deeper analysis.\"", "receiving_text": "—", "translation": "untranslated: composed as received (no archive claim in this lineage)", "archive_claims": [] }, { "lineage": "neighbours: the AI Incident Database", "field_claims": [ "F60" ], "received": "F60 (B5, Acknowledgements ¶2): \"Many experts provided invaluable feedback and suggestions, including Sean McGregor (Responsible AI Collaborative)\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-13, MMRS-16, MMRS-29, MMRS-15", "archive_claims": [ "MMRS-13", "MMRS-16", "MMRS-29", "MMRS-15" ] }, { "lineage": "holding: the Tow Center study", "field_claims": [ "F62", "F63", "F64" ], "received": "F62 (B7, item title and summary): \"AI Search Engines Fabricate News Citations, Undermining Publishers … A study by the Tow Center found that eight AI-powered search engines and chatbots, including ChatGPT, Perplexity and Elon Musk’s Grok3, frequently misquote or fabricate news citations, remixing publisher content without credit and driving down traffic. Over 60% of responses contained wrong or broken links, undermining journalistic credibility.[AI generated]\"; F63 (B7, label explanation; harm types): \"Hence, the event meets the criteria for an AI Incident as it involves realized harm caused by AI system outputs.[AI generated] … Harm types Economic/PropertyReputationalPublic interest\"; F64 (B7, articles list (25 articles, 2025-03-11 to 2025-03-17)): \"Articles about this incident or hazard … New study finds that AI search tools are 60 percent inaccurate on average 2025-03-11 TechSpot\"", "receiving_text": "—", "translation": "untranslated: composed as received beside MMRS-14, MMRS-28", "archive_claims": [ "MMRS-14", "MMRS-28" ] }, { "lineage": "reception: video", "field_claims": [ "F61" ], "received": "F61 (B6, oEmbed title (video, 29:48)): \"Establishing Effective Serious Incident Reporting Frameworks – Jimmy Farrell | IASEAI 2025\"", "receiving_text": "—", "translation": "untranslated: composed as received (no archive claim in this lineage)", "archive_claims": [] } ], "receiving_texts": "equal-terms case: every field claim composed as received, beside the archive claims it stands with" }, "measure": { "rule": "v0.9 §4.10: KO_B sentences surviving in the KO at similarity ≥ 0.95, with positions (computed as apply_v09.py does: difflib.SequenceMatcher ratio, the best KO sentence for each KO_B sentence)", "v0.9": { "surviving": 0, "of": 11, "positions": [ null, null, null, null, null, null, null, null, null, null, null ] }, "opening_run": { "rule": "v0.9 §5.10 (equal-terms case): the field-only KO's sentences as the KO's opening run in their own order", "breach": false }, "blocks": { "rule": "v0.9 §4.1 (equal-terms and mixed cases): the KO does not open with the field as a block or close with the archive as one; sentences typed by the ledgers of the claims they cite (F field-only, A archive-only, M both)", "sequence": "MMMFMMFFMMFFAFAMFFFAMFAFMMAFA", "field_only_head": 0, "archive_only_tail": 1, "breach": false }, "max_similarity": [ 0.146, 0.255, 0.215, 0.123, 0.113, 0.246, 0.128, 0.231, 0.138, 0.116, 0.071 ], "ko_sentence_1_max_similarity_to_KO_B": 0.231 }, "homonyms": { "names": [ "AIM (Google AI Mode, in the archive's own usage: #170, 'Google AI Mode (AIM), invoked where no Overview triggered')" ], "basis": "the MMRS record carries no disambiguation for this entity; the cards carry no homonym. The string pass found the acronym AIM in the archive only as Google AI Mode (#170, D0001–D0004), a collision on the acronym in the archive's own usage, set aside at reading (traversal reading.json). 'AI Incidents Monitor' is the entity's former name (F22), not a homonym. The neighbours (AI Incident Database, AIAAIC Repository), the operator's platform (the OECD.AI Policy Observatory), its data supplier (Event Registry), its supporter (the Patrick J. McGovern Foundation), the common reporting framework and the Tow Center study it holds are distinct entities, kept as relations, never senses of the name.", "distinct_entities": [ "ai-incident-database", "aiaaic-repository", "OECD.AI Policy Observatory", "Event Registry", "Patrick J. McGovern Foundation", "OECD common reporting framework for AI incidents", "Google AI Mode (AIM, #170)" ], "substitutes": { "T": [] }, "register": { "address_1": { "slug": "oecd-ai-incidents-and-hazards-monitor-aio-20261029", "obs_id": "OBS-23f50c1b7e04" } } }, "notes": [ "One address, Google AI Overview (expanded), signed out, incognito, 2026-10-29: oecd-ai-incidents-and-hazards-monitor-aio-20261029 (OBS-23f50c1b7e04, ADDR-de46fae4a143, six cards). A NEW address; the MMRS record landed the same night and no card cites it (capture q_kind).", "Wikipedia: no article for the monitor or for OECD.AI (raw wikitext 404 for seven titles; rendered 404; API: titles missing; Special:Search: no results; no redirect). The OECD article does not mention the monitor or OECD.AI. An API full-text search for 'OECD.AI' (not saved verbatim, so not ledgered) returned articles that cite the monitor's machine-written item pages as references ('OECD AI Monitor' in 'Emily Hart (virtual influencer)' and 'Artificial intelligence controversies'; items in 'Halupedia', 'Daisy Dixon', 'Google Chrome').", "Fetching: every reachable page saved verbatim with curl -sSL -m 30 -A 'Mozilla/5.0' on 2026-10-29 between 05:06:49 and 05:22:03 UTC, raw and text with sha256 in the field ledger. Card URLs were found by WebSearch on title and site; the video on YouTube's results page for the card title, confirmed by oEmbed. oecd.org HTML pages answer HTTP 403 (a Cloudflare challenge) to the workspace: B2 (the topic page 'AI risks and incidents') is unreachable, three requests, its snippet stands; B4's card page (the trends paper's HTML) answered 403 and the same publication was read as its PDF from oecd.org/content/dam (HTTP 200), linked from its OECD.AI page; B5 is a PDF (its title metadata is the card's title). The challenge was not circumvented.", "Own pages: the MMRS record's two sources (oecd.ai/en/incidents, oecd.ai/en/incidents-methodology) are cards B1 and B3, carried once, as cards; the holding oecd.ai/en/incidents/2025-03-11-8537 (B7) was not surfaced by T and is admitted beside the cards by the operator's ruling.", "The archive's deposits do not name the monitor by any name, its address or OECD.AI; the acronym AIM occurs in #170 as Google AI Mode (rejected, a collision recorded under homonyms). One deposit is admitted at O-class: #637 (The Encyclotron, 2026-04-08), 'Not media monitoring. Media monitors track coverage of events.' (O637-01), a distinction by contrast, borderline, admitted with its reason; its hop of 11 was read and rejected. The archive's bearing is carried by its own records: the MMRS record (MMRS-01–31) and the capture (CAP-1-01–08).", "The capture's open finding resolved by the field (reported, not fixed): CAP-1-06 ('FOURTEEN THEMES … Not verified here') is now readable against B4: the fourteen themes are the OECD trends paper's analysis (OECD AI Papers No. 53, February 2026), k-means clusters of the monitor's LLM-generated summaries from January 2022 to September 2025 consolidated from 25 to 14 (F35, F41–F43); the methodology page gives none; T composes them as the monitor's own 'Thematic Classification'.", "Defects and gaps found in the MMRS record, reported and not fixed. (1) census 'unchecked: launch date, licence': the field partly answers both — the expert group was formalised in January 2023 to develop the monitor (B5, F50), the monitor classified with 'traditional machine learning techniques' before November 2024 (B3, F34), and the trends paper's data run from January 2022 (B4, outside the ledgered spans); its use is subject to the OECD's terms and conditions (B3, F22); the trends paper, not the monitor's data, is CC BY 4.0 (F47). (2) census 'operator: OECD.AI' (basis operator, from the methodology page): the page names the OECD.AI expert group on AI incidents as initiator and developer with the support of the Patrick J. McGovern Foundation (F11); the record carries neither. (3) the record's account of selection ('An event enters when reputable outlets report it') does not carry the trends paper's 'top 50% of media outlets, based on the Alexa traffic rankings' (F39) or the methodology page's ranking of articles by their outlets' national ranking (F29). (4) the record's 'about 18,579' (observed 2026-10-28) and the page's 'About 18579' on 2026-10-29 are equal, while the methodology page states about 10 (F32) and about 30 (F33) new items a day: the count may be cached or the page's results line static; not settled. (5) the record names the three categories of the methodology page by implication ('sort them into incidents and hazards'); the monitor's own machine-written explanations also rule out 'Complementary Information' and 'Beneficial Use' (F9), classes the methodology page does not list.", "Field discrepancies, carried: the methodology page gives two daily yields, 'around 10' (F32) and 'on average 30' (F33), marked internally_inconsistent (open block); the methodology page lists three classes (F26) while the machine-written explanations name further ones (F9) (open block); the same video (kSuVPwr5Ll0) is 29:48 on YouTube's results page and on this card, while the Capture Registry's record of the same card at 'what counts as an incident in the ai incident database' (card 9) notes 'video, 19m' (reported, not fixed).", "The case and ρ (decided): equal, ρ = extends; the MMRS record describes the monitor from its own pages, with a basis per field, and its central reading (a machine reception of the news) draws the consequence of the field's own statements that the record is LLM-written (F28, F8, F41); the field holds the ontology, so the suppressed witness does not apply. No deposit defines the archive's entity; the defining material is the MMRS record and the capture record, with #637 at the class.", "Opening-run check (§5.10): KO sentence 1 is new (max similarity to any KO_B sentence 0.231); no KO_B sentence survives at ≥ 0.95 (highest 0.255); sentence types MMMFMMFFMMFFAFAMFFFAMFAFMMAFA (field-only head 0, archive-only tail 1).", "Not composed into the row, for the operator: the monitor's page states its data processing is 'powered by Microsoft Azure' (F6, composed), and the methodology page's '[AI generated]' practice extends to the label explanations ('Why's our monitor labelling this an incident or hazard?'), which are themselves machine compositions about news articles (F8): the record composes its own reasons." ], "composition_v09": { "source": "this row (composed directly; no rebuild/negative-of-the-negative/v09/oecd-ai-incidents-monitor.json)", "composer": "Claude (Opus 5.5), subagent, 2026-10-29", "defining_material_read": [ { "what": "MMRS record, census row", "file": "/home/claude/fleet/machinemediation-org/data/meta-aggregator/census.json (rows[slug == 'oecd-ai-incidents-monitor'], schema_note)", "read": "whole" }, { "what": "MMRS record, article", "file": "/home/claude/fleet/machinemediation-org/scripts/meta_aggregator/articles.py (ARTICLES['oecd-ai-incidents-monitor']; the sentences naming it in ARTICLES['ai-incident-database'] and ARTICLES['aiaaic-repository'])", "read": "whole" }, { "what": "MMRS record, rendered", "file": "/home/claude/fleet/machinemediation-org/aggregators/oecd-ai-incidents-monitor/index.html", "read": "text, for agreement with the sources" }, { "what": "capture record", "file": "data/EA-WG-CAPTURES-01.json (oecd-ai-incidents-and-hazards-monitor-aio-20261029)", "read": "transcript, transcript_raw, cite_list, d, reading, findings, analysis" }, { "dep": 637, "file": "data/texts/AXN-01ED-text.md", "read": "§VII–§IX around the admitted sentence" }, { "what": "the specification", "file": "data/texts/AXN-0700-text.md", "read": "§0–§11 and Appendix A whole" } ], "cuts": [], "notes": "equal case; KO 29 sentences, all 64 field and 40 archive claims cited; P body 350 words; KO_B 11 sentences, none surviving at ≥ 0.95" }, "field_by_origin": { "cards": [ "B1", "B2", "B3", "B4", "B5", "B6" ], "own_pages": [ "B7" ], "wikipedia": [ "W1", "W2" ], "unreachable": [ "B2" ], "title_only": [ "B6" ], "read_through_another_instance": [ "B4 (card page 403; the same publication read as its PDF)" ], "wikipedia_state": "no article: no English Wikipedia page exists for 'OECD AI Incidents and Hazards Monitor', 'OECD AI Incidents Monitor', 'AI Incidents Monitor', 'AI Incidents and Hazards Monitor', 'OECD.AI', 'OECD AI Policy Observatory' or 'OECD AI Principles' (raw wikitext HTTP 404; rendered page 404, 'Wikipedia does not have an article with this exact name.'; API query: all titles missing; Special:Search for the quoted names: no results), and no redirect; the OECD article (raw wikitext and rendered, HTTP 200) does not mention the monitor or OECD.AI. Several articles cite the monitor's item pages as references (an API search for 'OECD.AI', not saved)", "sources": [ { "id": "B1", "origin": "card", "role": "Overview card 1; also an own page the MMRS record cites (oecd.ai/en/incidents: count, self-description)", "card": "OECD AI Policy Observatory — OECD AI Incidents Monitor, an evidence base for trustworthy AI", "url": "https://oecd.ai/en/incidents", "kind": "Overview card (own page)", "surfaced_by_T": [ { "slug": "oecd-ai-incidents-and-hazards-monitor-aio-20261029", "card": 1 } ], "files": { "B1.html": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B1.html", "sha256": "e1681d645755e7c4094dd031225d55214520c0b6b0afdac1b6d6dc361430c679" }, "B1.txt": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B1.txt", "sha256": "74cf4b7530591ace479bca49d0af9594fbda19d94a0fe781eeb9aef049c13d7f" } }, "fetched": { "at": "2026-10-29T05:06:49Z", "http": 200, "url": "https://oecd.ai/en/incidents", "raw": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B1.html", "raw_sha256": "e1681d645755e7c4094dd031225d55214520c0b6b0afdac1b6d6dc361430c679", "text": [ "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B1.txt" ] }, "card_title": "OECD AI Incidents Monitor, an evidence base for trustworthy AI", "card_snippet": "Evolution of incidents and hazard by affected stakeholder As percentage of total AI events Note: An AI incident or hazard can be reported by one or more news ar...", "card_note": null, "names_operator": "yes ('OECD.AI' in the saved text)", "names_archive": "no (none of the Crimson Hexagonal Archive, Alexanarch, Lee Sharks, machinemediation, EA-WG-CAPTURES-01 in the saved text, card title and snippet)", "note": "the page title is the card's title; the request redirected to the default search view (search_terms=[], from_date=1900-10-29, to_date=2026-10-29, order_by=date, num_results=20); the card snippet's chart caption and note are on the page ('Evolution of incidents and hazard by affected stakeholder', 'As percentage of total AI events', the note F6). The result list is the day's (20 items, each with an '[AI generated]' summary and explanation); only the header, introduction, note, count and two explanations are ledgered" }, { "id": "B2", "origin": "card", "role": "Overview card 2", "card": "OECD — AI risks and incidents", "url": "https://www.oecd.org/en/topics/ai-risks-and-incidents.html", "kind": "Overview card", "surfaced_by_T": [ { "slug": "oecd-ai-incidents-and-hazards-monitor-aio-20261029", "card": 2 } ], "files": { "B2.html": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B2.html", "sha256": "54fda3deeb29a2f2dac931dfcb8951b91e2c95ffe18b43647dc8a6662c5b2a2d" } }, "fetched": { "at": "2026-10-29T05:08:47Z", "http": 403, "url": "https://www.oecd.org/en/topics/ai-risks-and-incidents.html", "raw": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B2.html", "raw_sha256": "54fda3deeb29a2f2dac931dfcb8951b91e2c95ffe18b43647dc8a6662c5b2a2d", "text": [] }, "unreachable": true, "card_title": "AI risks and incidents", "card_snippet": "All actors must monitor AI incidents and hazards to mitigate risks As AI adoption continues to grow, successful risk mitigation will require a solid evidence ba...", "card_note": null, "names_operator": "no ('OECD.AI' absent from the card title and snippet; page unread)", "names_archive": "no (card title and snippet; page unread)", "note": "unreachable: URL found by WebSearch (title and site: oecd.org/en/topics/ai-risks-and-incidents); HTTP 403 (a Cloudflare 'Just a moment...' challenge page) on three requests 2026-10-29 between 05:07 and 05:08:47Z (.html, /sub-issues/ and extensionless paths); no further retry, the challenge not circumvented; the card snippet stands as its only text. oecd.ai/en/site/incidents (fetched, not saved as a source) is a different page and does not carry the snippet" }, { "id": "B3", "origin": "card", "role": "Overview card 3; also an own page the MMRS record cites (oecd.ai/en/incidents-methodology: method, definitions)", "card": "OECD AI Policy Observatory — Overview and methodology of the AI Incidents and Hazards Monitor - OECD.AI", "url": "https://oecd.ai/en/incidents-methodology", "kind": "Overview card (own page)", "surfaced_by_T": [ { "slug": "oecd-ai-incidents-and-hazards-monitor-aio-20261029", "card": 3 } ], "files": { "B3.html": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B3.html", "sha256": "26400e79e67de5eeed8d471a5ddfc5e61a71156f0d8614bade6dd6f6ad18b066" }, "B3.txt": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B3.txt", "sha256": "edfcba595f3fb6b8c8696c92b6ea402d8a9c0c8d323dce34b1ed4894606df5cc" } }, "fetched": { "at": "2026-10-29T05:06:50Z", "http": 200, "url": "https://oecd.ai/en/incidents-methodology", "raw": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B3.html", "raw_sha256": "26400e79e67de5eeed8d471a5ddfc5e61a71156f0d8614bade6dd6f6ad18b066", "text": [ "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B3.txt" ] }, "card_title": "Overview and methodology of the AI Incidents and Hazards Monitor - OECD.AI", "card_snippet": "Overview The AI Incidents and Hazards Monitor (AIM) was initiated and is being developed by the OECD.AI expert group on AI incidents with the support of the Pat...", "card_note": null, "names_operator": "yes ('OECD.AI' in the saved text)", "names_archive": "no (none of the Crimson Hexagonal Archive, Alexanarch, Lee Sharks, machinemediation, EA-WG-CAPTURES-01 in the saved text, card title and snippet)", "note": "the card snippet is the page's Overview ¶1 (F11); the page gives no count of themes (CAP-1-06 confirmed); its summary table gives two daily rates (F32 about 10, F33 about 30); 'Please find this archive' (the pre-November-2024 method) links an archive not fetched" }, { "id": "B4", "origin": "card", "role": "Overview card 4", "card": "OECD — Trends in AI incidents and hazards reported by the media | OECD", "url": "https://www.oecd.org/en/publications/trends-in-ai-incidents-and-hazards-reported-by-the-media_4f5ff43c-en.html", "kind": "Overview card (OECD Artificial Intelligence Papers No. 53, February 2026; read as its PDF)", "surfaced_by_T": [ { "slug": "oecd-ai-incidents-and-hazards-monitor-aio-20261029", "card": 4 } ], "files": { "B4-html-403.html": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B4-html-403.html", "sha256": "0bbd53ffd919f2ffebdf71abe9454306a939b1ed9b18cce4911898ce3c482208" }, "B4.pdf": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B4.pdf", "sha256": "aab2727da192624f766ed676a616b3de376ed92430ad3826063e291d951d1ce5" }, "B4.pdf.txt": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B4.pdf.txt", "sha256": "ec4d42002736b0233c5c4def2222bc2fece78a8031274c9b8e3eacc410d0467d" } }, "fetched": { "at": "2026-10-29T05:08:47Z", "http": 200, "url": "https://www.oecd.org/content/dam/oecd/en/publications/reports/2026/02/trends-in-ai-incidents-and-hazards-reported-by-the-media_7c824ca9/4f5ff43c-en.pdf", "raw": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B4.pdf", "raw_sha256": "aab2727da192624f766ed676a616b3de376ed92430ad3826063e291d951d1ce5", "text": [ "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B4.pdf.txt" ] }, "card_title": "Trends in AI incidents and hazards reported by the media | OECD", "card_snippet": "8 The OECD.AI Incidents and Hazards Monitor (AIM) tracks these media-reported AI incidents and hazards in real-time and leverages the OECD common reporting fram...", "card_note": null, "names_operator": "yes ('OECD.AI' in the saved text)", "names_archive": "no (none of the Crimson Hexagonal Archive, Alexanarch, Lee Sharks, machinemediation, EA-WG-CAPTURES-01 in the saved text, card title and snippet)", "note": "the card's title is the publication's HTML page on oecd.org, which answered HTTP 403 (Cloudflare challenge) at 05:08:47Z (B4-html-403.html; also oecd-ilibrary.org and read.oecd-ilibrary.org, 403); the same publication's PDF, linked from its OECD.AI page (oecd.ai/en/ai-publications/trends-in-ai-incidents-and-hazards-reported-by-the-media), was fetched from oecd.org (HTTP 200, application/pdf, 36 pages; PDF title 'Trends in AI incidents and hazards reported by the media (EN)', DOI 10.1787/4f5ff43c-en) and its text extracted with pdftotext. The card's snippet is the PDF's executive summary (page 8; F38). The paper is CC BY 4.0 (F47); quoted in short spans" }, { "id": "B5", "origin": "card", "role": "Overview card 5 (the OECD's May 2024 definitions report, named in the task)", "card": "OECD — Defining AI incidents and related terms (EN)", "url": "https://www.oecd.org/content/dam/oecd/en/publications/reports/2024/05/defining-ai-incidents-and-related-terms_88d089ec/d1a8d965-en.pdf", "kind": "Overview card (OECD Artificial Intelligence Papers No. 16, May 2024; PDF)", "surfaced_by_T": [ { "slug": "oecd-ai-incidents-and-hazards-monitor-aio-20261029", "card": 5 } ], "files": { "B5.pdf": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B5.pdf", "sha256": "9c46e00f1b3ff6de1aeb99564df9e57cc53b60db50e412c8048b213836af45e1" }, "B5.pdf.txt": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B5.pdf.txt", "sha256": "97db457717a855e24cca3ab852d7e47fe71ff40a3aecdc8a8c4b86358c880f87" } }, "fetched": { "at": "2026-10-29T05:08:48Z", "http": 200, "url": "https://www.oecd.org/content/dam/oecd/en/publications/reports/2024/05/defining-ai-incidents-and-related-terms_88d089ec/d1a8d965-en.pdf", "raw": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B5.pdf", "raw_sha256": "9c46e00f1b3ff6de1aeb99564df9e57cc53b60db50e412c8048b213836af45e1", "text": [ "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B5.pdf.txt" ] }, "card_title": "Defining AI incidents and related terms (EN)", "card_snippet": "OECD May 2024 Report No. 16 defines AI incident terminology for reporting frameworks and the AI Incidents Monitor (AIM), approved 14 Mar 2024 by the Digital Pol...", "card_note": null, "names_operator": "yes ('OECD.AI' in the saved text)", "names_archive": "no (none of the Crimson Hexagonal Archive, Alexanarch, Lee Sharks, machinemediation, EA-WG-CAPTURES-01 in the saved text, card title and snippet)", "note": "the card's title is the PDF's own title metadata ('Defining AI incidents and related terms (EN)', DOI 10.1787/d1a8d965-en), so the card is the PDF; the URL is the PDF link on the paper's OECD.AI page (oecd.ai/en/ai-publications/defining-ai-incidents-and-related-terms); the publication's HTML page on oecd.org answered 403. The card snippet is a summary ('defines AI incident terminology … approved 14 Mar 2024 by the Digital Pol...'), its facts in F48–F49. 15 pages; text by pdftotext" }, { "id": "B6", "origin": "card", "role": "Overview card 6 (video, 29:48)", "card": "YouTube·International Association for Safe & Ethical AI — Establishing Effective Serious Incident Reporting Frameworks – Jimmy Farrell", "url": "https://www.youtube.com/watch?v=kSuVPwr5Ll0", "kind": "Overview card (video)", "surfaced_by_T": [ { "slug": "oecd-ai-incidents-and-hazards-monitor-aio-20261029", "card": 6 } ], "files": { "B6.json": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B6.json", "sha256": "5a063beb92dbc8496affd425301302f30259d11aa5d4aa0b2cca557e0e8d57f5" }, "B6.txt": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B6.txt", "sha256": "8c9f8a79bff55d6b7979d200bdd7d2a6752644483f88c701df0d80618d31906c" }, "B6-search.html": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B6-search.html", "sha256": "0ef80d04d98bf92b8980a04b4ed318131ef2a9f0465499c80167cac234ba940b" } }, "fetched": { "at": "2026-10-29T05:09:49Z", "http": 200, "url": "https://www.youtube.com/oembed?url=https://www.youtube.com/watch?v=kSuVPwr5Ll0&format=json", "raw": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B6.json", "raw_sha256": "5a063beb92dbc8496affd425301302f30259d11aa5d4aa0b2cca557e0e8d57f5", "text": [ "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B6.txt" ] }, "card_title": "Establishing Effective Serious Incident Reporting Frameworks – Jimmy Farrell ...", "card_snippet": null, "card_note": "video, 29:48", "names_operator": "no ('OECD.AI' absent from the oEmbed title and the card)", "names_archive": "no (oEmbed title and card)", "note": "located on YouTube's results page for the card title (B6-search.html, 05:09:3xZ: kSuVPwr5Ll0, length 29:48, the card's duration); oEmbed title adds '| IASEAI 2025'; no transcript fetched; admitted by title (§3.1(a)). The same video is the AI Incident Database row's B14; the Capture Registry's record of that card (what-counts-as-an-incident-in-the-ai-incident-database-aio-20261029, card 9) notes it 'video, 19m'" }, { "id": "B7", "origin": "own_page", "role": "own page the MMRS record cites (relation 'holding'); not surfaced by T", "card": "OECD AI Policy Observatory — AI Search Engines Fabricate News Citations, Undermining Publishers (AIM item 2025-03-11-8537)", "url": "https://oecd.ai/en/incidents/2025-03-11-8537", "kind": "own page (an item of the monitor)", "surfaced_by_T": "no (not surfaced by T; admitted beside the cards by the operator's ruling of 2026-10-29)", "files": { "B7.html": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B7.html", "sha256": "9edcbeb65fe222f43523c5d6fc97202e09ccb3fd3651b9a23b2ccd7b3ed5915c" }, "B7.txt": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B7.txt", "sha256": "4dc98555510507f272be5da32e42468f466797e0aa45fd7d1a8cfd035a65d05c" } }, "fetched": { "at": "2026-10-29T05:08:49Z", "http": 200, "url": "https://oecd.ai/en/incidents/2025-03-11-8537", "raw": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B7.html", "raw_sha256": "9edcbeb65fe222f43523c5d6fc97202e09ccb3fd3651b9a23b2ccd7b3ed5915c", "text": [ "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/B7.txt" ] }, "names_operator": "yes ('OECD.AI' in the saved text)", "names_archive": "no (none of the Crimson Hexagonal Archive, Alexanarch, Lee Sharks, machinemediation, EA-WG-CAPTURES-01 in the saved text, card title and snippet)", "note": "the item page: title, '[AI generated]' summary and label explanation, principles, industries, stakeholders, harm types, 25 articles (2025-03-11 to 2025-03-17), each with its own label explanation link. The MMRS record's own-page sources (oecd.ai/en/incidents-methodology, oecd.ai/en/incidents) are cards B3 and B1, carried once, as cards" }, { "id": "W1", "origin": "wikipedia", "result": "no article", "checked": "2026-10-29 05:09:57–05:22:03 UTC", "requests": [ { "url": "https://en.wikipedia.org/w/index.php?title=OECD_AI_Incidents_and_Hazards_Monitor&action=raw&redirect=no", "http": 404 }, { "url": "https://en.wikipedia.org/w/index.php?title=OECD_AI_Incidents_Monitor&action=raw", "http": 404 }, { "url": "https://en.wikipedia.org/w/index.php?title=AI_Incidents_Monitor&action=raw", "http": 404 }, { "url": "https://en.wikipedia.org/w/index.php?title=AI_Incidents_and_Hazards_Monitor&action=raw&redirect=no", "http": 404 }, { "url": "https://en.wikipedia.org/w/index.php?title=OECD.AI&action=raw", "http": 404 }, { "url": "https://en.wikipedia.org/w/index.php?title=OECD_AI_Policy_Observatory&action=raw", "http": 404 }, { "url": "https://en.wikipedia.org/w/index.php?title=OECD_AI_Principles&action=raw&redirect=no", "http": 404 }, { "url": "https://en.wikipedia.org/wiki/OECD_AI_Incidents_and_Hazards_Monitor", "http": 404, "file": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/W1.html", "sha256": "b373c35b4f50e7dbf4194f3fe68ec39776f5013f9eb221d6fca1828ed1863aa1", "says": "Wikipedia does not have an article with this exact name." }, { "url": "https://en.wikipedia.org/w/api.php?action=query&format=json&titles=OECD_AI_Incidents_and_Hazards_Monitor|OECD_AI_Incidents_Monitor|AI_Incidents_Monitor|AI_Incidents_and_Hazards_Monitor|OECD.AI|OECD_AI_Policy_Observatory|OECD_AI_Principles|AI_Incident_Monitor|OECD_AI_Incident_Monitor|AIM_(OECD)", "http": 200, "file": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/W1.api.json", "sha256": "c93f211697c66593a7a5fb233e2d02603098fd958505a66874f313d7825c8aca", "says": "all ten titles 'missing' (two earlier attempts answered 429, rate-limited)" }, { "url": "https://en.wikipedia.org/w/index.php?search=%22AI+Incidents+Monitor%22&title=Special%3ASearch&profile=advanced&fulltext=1&ns0=1", "http": 200, "file": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/W1-search.html", "sha256": "b0afefd8289d609531dc132d24047e0c445b3af3fee6f9144d09e9ea3036960a", "says": "There were no results matching the query." }, { "url": "https://en.wikipedia.org/w/index.php?search=%22OECD+AI+Incidents%22&title=Special%3ASearch&profile=advanced&fulltext=1&ns0=1", "http": 200, "file": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/W1-search2.html", "sha256": "1864aa0f81d0d902e3ea8851111b362f4d621872c5d950850a5860ee9dfe4973", "says": "There were no results matching the query." } ], "redirect": "none (no title variant resolves; no redirect page)", "note": "No article for the monitor, for OECD.AI or for the OECD AI Policy Observatory exists on English Wikipedia on 2026-10-29. A full-text API search for 'OECD.AI' (05:09:57Z, not saved; 18 hits) returned only articles citing oecd.ai pages as references, several of them items of the monitor (e.g. 'OECD AI Monitor' as the work cited in 'Emily Hart (virtual influencer)' and 'Artificial intelligence controversies'; oecd.ai items in 'Halupedia', 'Daisy Dixon', 'Google Chrome'): the encyclopedia takes the monitor's machine-written item pages as sources and has no page for the monitor. Not ledgered (the search response was not saved verbatim)." }, { "id": "W2", "origin": "wikipedia", "result": "the OECD article: no mention of the monitor", "url": "https://en.wikipedia.org/w/index.php?title=OECD&action=raw", "http": 200, "files": { "W2.wiki": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/W2.wiki", "sha256": "03ca69ee493875ab72e364695ebebf19fc17a7e980e8cf2d1a4b5bd7c799744e" }, "W2.html": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/W2.html", "sha256": "0f6759444535f525951c5fe2351c9027debb843eaeaf83b8d5a3ffddc00f836b" }, "W2.txt": { "path": "datasets/negative-of-the-negative/v2/intake/oecd-ai-incidents-monitor-20261029/W2.txt", "sha256": "4d938f4bb993922e7e9f839a4ef4cd04a8f30c10e3d9d4cf0f6da635d9250105" } }, "fetched_at": "2026-10-29T05:10:57Z", "note": "raw wikitext (85,737 bytes) and the rendered article (https://en.wikipedia.org/wiki/OECD, HTTP 200): no occurrence of 'incident', 'OECD.AI', 'AI Policy Observatory' or 'AI Principles'; artificial intelligence appears once, in the OECD Observer paragraph (its last issue, 2019, 'looking ahead at artificial intelligence'). The OECD's article does not mention the monitor or OECD.AI." } ] } }