C2PA Conformance Program and Explorer

register entry c2pa-conformance-program-aio-20261029 OBS-47bccd9af209 · file as first run sha256 169f414acf8186da…

• T1-1 definition: 'a risk-based governance and verification framework run by the Coalition for Content Provenance and Authenticity to ensure hardware and software products properly implement the Content Credentials specification' (markers 1, 2)
• T1-2 conforming products list: 'Verified generator and validator products are publicly listed on the C2PA Conformance Explorer.'
• T1-3 generator products: 'Software or hardware tools that create asset manifests, apply assertions, and sign content using X.509 certificates (categorized by Level 1 or Level 2 assurance levels).'
• T1-4 validator products: 'Applications or services that read, check, and verify manifest stores and cryptographic signatures.'
• T1-5 certificate authorities: 'Entities that issue claim-signing certificates conforming to C2PA policy and the official Trust List.' (markers 1–5)
• T1-6 evaluation process: 'Applicants submit an expression of interest, sign legal agreements, provide technical evidence, and run self-test tools against program requirements.' (markers 1, 2)
• T1-7 trust and security: 'Only conforming generator products can receive production claim-signing certificates from approved CAs, creating end-to-end tamper evidence for digital media.' (markers 1, 2)
• T1-8 closer: 'If you want, I can help you:' and three offers (look up vendors on the list; explain Level 1 and Level 2; detail how to submit a product)

• B1 Coalition for Content Provenance and Authenticity (C2PA) — Conformance - C2PA card; copied from the census (c2pa-conformance--conformance), HTTP 200, sha256 2478cf522501d050
• B2 Content Authenticity Initiative — C2PA conformance program | Open-source tools for content authenticity and provenance card; HTTP 200, B2.html sha256 a77e497ed496753c
• B3 Content Authenticity Initiative — Raising the bar for trust: introducing the C2PA Conformance Program card; HTTP 200, B3.html sha256 957e25fbaccbee4e
• B4 C2PA Specifications — C2PA Conformance Explorer card; HTTP 200, B4.html sha256 1b8830b8b9a126b3; text from the script bundle (B4-main-3SWFN3GG.js)
• B5 SSL.com — How to Submit to the C2PA Conformance Program card; HTTP 200, B5.html sha256 c6f361ab501b1692
• B6 Coalition for Content Provenance and Authenticity (C2PA) — C2PA FAQ card; HTTP 200, B6.html sha256 2895682fb4ed4f34
• B7 trufo.ai — C2PA Conformance Program - Trufo card; HTTP 200, B7.pdf sha256 d1d4b93232086647; pdftotext -layout
• B8 LinkedIn·Scott Perry — C2PA Conformance Program Documents Released v0.2 Upgrade - LinkedIn unreachable: URL not found (six searches); card title and snippet only
• B9 C2PA Specifications — C2PA and Content Credentials Explainer 2.4 own_page; copied from the census (c2pa-conformance--explainer-2.4), HTTP 200, sha256 e1872450e1ae7f6a
• B10 c2pa-org/conformance-public — conforming-products-list.json own_page; copied from the census (c2pa-conformance--products-json), HTTP 200, sha256 e2fb3928fbb2ba3d
• B11 Wikipedia — C2PA Conformance Program (no article) Wikipedia; raw HTTP 404, rendered HTTP 404
• B12 Wikipedia — Content Authenticity Initiative § Coalition for Content Provenance and Authenticity Wikipedia; raw HTTP 200, rendered HTTP 200; rendered sha256 2665c40cc843a5c5
• B13 Wikipedia — Content Credentials § C2PA Conformance Program Wikipedia; raw HTTP 200, rendered HTTP 200; rendered sha256 5bfabcbc373de072

E_A, the archive's entity: the same program, in the provenance census record and the archive's deposits: a registry of the media lineage that expresses and records provenance and tests no survival, its object a digital asset with its Content Credential and its unit the product or signer (census record); the governance of the artifact dimension of provenance, which verifies this file, this source, this time, and the signing event, not the truth of what is signed (#725), not authorship (#729), the camera's testimony rather than the author's, by trust chains dependent on certificate authorities and revocable (#1444), on a trust model of assumption and opt-in in which 'trusted does not necessarily imply trustworthy' (#1443); with no reach into text, synthesis or the composition layer (#725, #615, #127), and met by the archive at its own artifacts' Trust-List signing (#1409) PE-01 PE-04 PE-03 PE-05 PE-06 PE-10 D725-06 D725-08 D725-11 D729-04 D1444-01 D1443-01 D725-05 D615-01 D127-02 R1409-02 CAP-1-07; defining papers #725, #729, #1443, #1444, #1409, #615, #127

E_F, the field's entity (type C — a conventional reading holds it (public entity: a conformance program and its registry of conforming products and trust lists, run by a standards coalition)): the Coalition for Content Provenance and Authenticity's risk-based governance program, administered by its Conformance Task Force under the Steering Committee, that designates generator and validator products as conforming to the Content Credentials specification and certification authorities as adhering to its Certificate Policy, publishes the Conforming Products List, the Trust List and the TSA Trust List (browsed in the Conformance Explorer; 286 records on 2026-10-29), and admits to production claim-signing certificates only conforming generator products; opened June 2025, v0.2 since 31 July 2026; by its own statements it certifies correct production and validation of C2PA data, not the truth of provenance, not identity or attribution, and complete conformance only eventually

ρ = extends: the archive reads E_F as the entity and adds senses, mechanisms and limits on the field's own object: the artifact dimension among independent dimensions (D725-01, D725-05, D729-01, O1386-01), what the signature establishes and the signed-forgery mechanism (D725-06, D725-08, D725-09, D725-11), the trust model and its revocability (D1443-01, D1443-03, D1444-01), authorship outside it (D729-03, D729-04), what the artifact layer does not reach (D615-01, D127-02, D621-01), the census's reading of what it carries (PE-05, PE-10) and what the composition leaves out (CAP-1-04, CAP-1-07) D725-01 D725-05 D729-01 O1386-01 D725-06 D725-08 D725-09 D725-11 D1443-01 D1443-03 D1444-01 D729-03 D729-04 D615-01 D127-02 D621-01 PE-05 PE-10 CAP-1-04 CAP-1-07. Guard: ρ = extends claims an addition and no more: #725's signed-forgery case is 'not a flaw of C2PA' but a property of all signature-based systems, and its own text says the existing frameworks 'were not designed for the dimension proposed here'; the field states the governing limit itself (F48, F50, F52), so the archive's senses extend a limit the operator already draws; #1443's trust-model reading is attributed to C2PA's own literature and its comparison is 'complementary rather than competing'; #127's and #1003's contrary statements on authorship are composed beside the operator's under P_coh.

The case: equal terms, ontology included (v0.9). The archive reads the field's entity as the entity: the census record describes the program from its own pages, each field with a verbatim quote and its basis, and the deposits that bear on it (#725, #729, #1443, #1444, #1409) take C2PA's signing and trust chain as given and place them, adding senses, mechanisms and limits on the field's own object: the artifact dimension of provenance among independent dimensions, what a signature establishes and does not (the signed-forgery mechanism), the trust model read from C2PA's own literature, the chain's revocability, authorship as outside it, and the reach the artifact layer lacks. #725 states its relation in those terms ('Extension and completion, not substitution or critique'; 'This is not a flaw of C2PA'), and #1443 holds the two 'complementary rather than competing'. The field itself states the governing limit (Explainer 2.4: no judgement of truth; no identity) that the archive's senses extend. No defining paper contests the field's ontology or supplies one the field lacks: ρ = extends, the equal-terms case (§0.9, §1.2a). Two readings are held under P_coh rather than ruled: #127's and #1003's against the operator's statement that the core specification does not support attribution, and, within the field, the FAQ's 'infeasible' forgery against the reported demonstrations.

#725 packet header, Relation: **Relation:** Extension and completion, not substitution or critique

#725 §0 ¶7: It is offered as a constructive extension of existing frameworks — not a critique of their adequacy in their own domains.

#725 §3.6, Correction: **Correction:** This is not a flaw of C2PA.

#725 §1.4 ¶4: The packet's claim is not that the existing frameworks fail. It is that they were not designed for the dimension proposed here.

#1443 §6.1 ¶6: These are **complementary rather than competing**, and saying so is stronger than claiming superiority

#729 §8.7: **C2PA prevents forgery; it does not constitute authorship.**

#provenance census census.json organizing_question: What provenance does this system carry, through which operation, on whose assertion, with what evidence of survival?

#provenance census census.json rows[slug == 'c2pa-conformance'], retention_results: not tested here: conformance concerns producing and validating C2PA data correctly

equal-terms case: every field claim composed as received, beside the archive claims it stands with. Reading the received column recovers the field (§5.9).

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 0 of 12 (positions [None, None, None, None, None, None, None, None, None, None, None, None]; composed first under v0.9, so there is no v0.7 object to measure against). Opening-run check (§5.10): no breach.

The C2PA Conformance Program gives assurance that products adhere to the Content Credentials specification and meet security requirements, producing and validating C2PA data correctly; it is a risk-based governance program designating generator and validator products as conforming and certification authorities as adhering to the Certificate Policy (c2pa.org/conformance; program document v0.2). It sits in the C2PA Governance Framework under the Steering Committee and is administered by the Conformance Task Force, set up in 2024 and co-chaired from the CAI, Arm and Google, through an independent Administrator (c2pa.org/conformance; Wikipedia; CAI blog; v0.2); the coalition, founded in 2021, has no article of its own on Wikipedia (Wikipedia). Conforming products go on a public list, the canonical Conforming Products List, where the generator product is always the signer and accountable for its assets, and the FAQ and SSL.com read a listing as visibility of trusted, interoperable and secure implementations, subject to removal or revocation (c2pa.org/conformance; v0.2; CAI docs; C2PA FAQ; SSL.com). Only conforming generator products may receive claim-signing certificates from CAs on the curated C2PA Trust List, signing claims with X.509 certificates, in what the CAI calls a unified system of standard, tools, conformance and trust model (v0.2; C2PA FAQ; CAI docs; SSL.com; CAI blog). There are two assurance levels, level 2 requiring hardware-backed keys and dynamic evidence, an instance's level possibly below the product's maximum; Google's Pixel 10 was first at the highest level (CAI docs; v0.2; SSL.com; CAI blog; Conforming Products List). Applicants sign a legal agreement, provide evidence, pass a technical evaluation and receive a Conformance Letter before a production certificate (CAI docs; SSL.com). It soft-launched in June 2025 and launched fully in October 2025; v0.1 ran from 2 June 2025 to 9 October 2026 and v0.2 began on 31 July 2026, supporting specifications 2.2 and 2.4 (CAI blog; Wikipedia; v0.2; Scott Perry; c2pa.org/conformance). The Conformance Explorer browses the live Conforming Products List, Trust List and TSA Trust List; the list held 286 conformant records, against Wikipedia's count of over 200 products and 10 organizations both conforming and on the Trust List (v0.2; Conformance Explorer; Conforming Products List; Wikipedia). What is signed is a Content Credential, a cryptographically bound, tamper-evident structure recording an asset's provenance, not DRM (Explainer 2.4; C2PA FAQ). Content Credentials do not judge whether provenance is true, only whether it is well-formed, untampered, valid and trusted; the core specification does not support attribution or address identity; provenance can be incomplete, removed or separated, and ingredients cannot be fully verified (Explainer 2.4; C2PA FAQ; Wikipedia). The program will only eventually require complete conformance to the specification's normative requirements (v0.2). The FAQ calls faking credentials infeasible, while Wikipedia reports demonstrated breaks in 2026 and criticism of bypasses; no dedicated camera or desktop application had achieved compliance (C2PA FAQ; Wikipedia).

The C2PA Conformance Program is the Coalition for Content Provenance and Authenticity's program of assurance "that products adhere to the Content Credentials specification, and fulfill a set of security requirements to ensure they are producing and validating C2PA data correctly" (c2pa.org/conformance); its own document defines it as "A risk-based governance program" whose processes, policies and requirements designate generator and validator products as conforming products and certification authorities as adhering to the C2PA Certificate Policy (C2PA Conformance Program v0.2, §2.11), and the provenance census records it, with its Explorer, as a registry of the media lineage that expresses and records provenance (provenance census record). The conformance page calls it "a risk-based, transparent and unbiased governance process" holding generator products, validator products and certification authorities accountable to the specification, the Certificate Policy and the Security Requirements (c2pa.org/conformance), the Content Authenticity Initiative's documentation lists the same three roles (CAI open-source docs), and Andy Parsons of the CAI likens it to the Bluetooth and Wi-Fi programs, "baseline assurance for content provenance" that validates an implementation "correctly creates, signs, and reads C2PA manifests" (CAI blog, 9 October 2025); the program document says why it exists: the specification "is limited without a robust governance framework and conformance program" giving relying parties transparency and governed parties accountability (v0.2, §1). It is part of a C2PA Governance Framework governed by the C2PA Steering Committee and administered by the C2PA Conformance Task Force (c2pa.org/conformance), a task force established in early-to-mid 2024 on the Trust over IP Governance Metamodel (Wikipedia, Content Credentials) and co-chaired by Parsons, Yogesh Deshpande (Arm) and Sherif Hanna (Google) (CAI blog); the program document has it "administered by an independent Administrator who is free from conflicts of interest" (v0.2, §6.6.3), and the census records its operator as the C2PA's Conformance Task Force, dependent on the Steering Committee, the Trust List, the TSA Trust List and the Certificate Policy (provenance census record); the coalition itself has no Wikipedia article of its own, its names redirecting to a section of the Content Authenticity Initiative article that gives it as "an open, cross-industry technical standards body established on February 22, 2021" (Wikipedia). Its unit is the listed product: "Conforming products are placed on a publicly-accessible list," which the operator says conveys confidence and "guarantees interoperability" (c2pa.org/conformance); the C2PA Conforming Products List is "The canonical record of all Conforming Products," and because the generator product "is always the Signer" and always the listed entity, it is accountable for the conformance of the assets it generates (v0.2, §2.12, §2.26); the census gives the unit as "a conforming generator or validator product, or a trust-list entry" and the provenance object as a digital asset with its Content Credential, where "the registry's own unit is the product or signer" (provenance census record). Acceptance adds a product to the list, which "indicates it is compliant," and a conforming validator "produces correct validation results" (CAI open-source docs); the FAQ says listed products are "vetted by the Administering Authority," "providing visibility into which implementations are trusted," and that content signed by them can be expected to carry "a higher level of implementation trust" (C2PA FAQ); SSL.com says a listing "signals to the broader ecosystem that your implementation is trustworthy, interoperable, and secure" (SSL.com, 16 July 2026); and the C2PA reserves the right to remove applicants from the trust lists or revoke conformant status (v0.2, §6.7.3). The program governs a chain of signing: "Only instances of conforming Generator Products are eligible to receive C2PA Claim Signing Certificates from a Certification Authority on the C2PA Trust List" (v0.2, §2.23); the Trust List is "a curated list of Certification Authorities" authorised to issue signing certificates only to products the program approved (C2PA FAQ); CAs that meet the certificate policy can be on it, and a conforming generator signs a claim "using a valid X.509 certificate on the C2PA trust list" (CAI open-source docs); CAs on it, SSL.com among them, "can only issue production-grade Claim Signing Certificates to conforming products" (SSL.com). The census names the relations the chain carries as signing, creation and edit history (ingredients) and trust-list membership, and its carrier as "signed C2PA manifests; the conformance lists as JSON" (provenance census record); Parsons calls the standard, the tools, the conformance process and the trust model together "a unified system for establishing digital provenance at scale" (CAI blog). #1443 reads the trust model the program governs from C2PA's own literature: trust "extended to many entities by assumption and opt-in," "validation is not required by the specification," coverage of the whole file not required, and "trusted does not necessarily imply trustworthy"; "C2PA attests what a device reports having done," which "requires trusting a chain of signers," where a content-derived identifier requires trusting SHA-256 (#1443, §6.1). #1443 holds the two "complementary rather than competing," a C2PA-signed capture that is also content-addressed and publicly witnessed being "better evidenced than either alone," and calls P.Sapph.Obbink "the human C2PA": "trust extended by assumption to a chain of signers one link deep; validation optional and unperformed" (#1443, §6.1); #1409 sets the anchors side by side, C2PA "in device and issuer certificates," the AXN regime "in content derivation plus an archive's public ledger" (#1409). #1444 adds that C2PA is "structurally dependent on certificate authorities and platform adoption," its "trust chains revocable by exactly the class of actor that issues terminations" (#1444), the power of removal and revocation the program document reserves to the C2PA (v0.2, §6.7.3). Two assurance levels are in operation: a product's level "indicates the level of confidence that claims it signs reflect its intended behavior," an instance's certificate may carry a level below the product's maximum (CAI open-source docs), and relying parties "are advised to review and understand the implications" of a level (v0.2, §5); Level 2 "demands more rigorous security controls, including hardware-backed key storage and dynamic security evidence," while SSL.com's rule of thumb names level 1 for devices and "at least level 1" for cloud services (SSL.com); Google's Pixel 10 was the first device at the highest level (CAI blog), and the list's first record, Google's Pixel Camera, carries maximum level 2 by Android key attestation (Conforming Products List). Applicants sign a legal agreement with the C2PA, provide evidence and work with the program's staff (CAI open-source docs); the technical evaluation checks that manifests are properly formatted and certificates handled per the Certificate Policy, and a Conformance Letter makes the product eligible for a production claim-signing certificate from SSL.com or any approved CA on the Trust List (SSL.com). The program was soft-launched in June 2025 at the Content Authenticity Summit at Cornell Tech and was "fully up and running" by October 2025 (CAI blog), Wikipedia giving 4 June 2025, a full public launch in October 2025 and the Interim Trust List deprecated on 31 December 2025 (Wikipedia, Content Credentials); its version table runs v0.1 from 2025-06-02 to its sunset on 2026-10-09, supporting specification v2.2, and v0.2 from 2026-07-31, sunset not set, supporting v2.2 and v2.4 (v0.2, Table 1), the document dated "Version 0.2, 2026-07-31" (v0.2, title page) and announced the same day on LinkedIn (Scott Perry, title and snippet only); the conformance page says it opened in June 2025 and is now v0.2 (c2pa.org/conformance), and the census records it active, "program opened June 2025, now v0.2," read as program v0.2 and Explainer 2.4 on 2026-10-29 (provenance census record). The Conformance Explorer is the program's "human-readable form of our public trust lists" (v0.2, §6.6.8), where one may "Browse, search, and filter the live versions" of the Conforming Products List, the Trust List and the TSA Trust List, each product record giving its program version, maximum assurance level and supported media (Conformance Explorer); the census gives access as "Conformance Explorer; JSON lists on GitHub" (provenance census record). The JSON list held 286 records when the census read it (provenance census record; Capture Registry, 2026-10-29), every record conformant, 254 of them generator and 32 validator products, 11 generator products at maximum level 2 (Conforming Products List); Wikipedia's section says the Explorer indexes "over 200 active products" while 10 organizations have deployed products that both complete the program and are on the Trust List (Wikipedia, Content Credentials). What a conforming product signs is a Content Credential: provenance, "as C2PA defines it," is "the facts about the history of a piece of digital content," and the Content Credential is "a cryptographically bound structure that records an asset’s provenance" (Explainer 2.4); the FAQ gives "tamper-evident, cryptographically signed data structures that travel with the asset" and says they are "not a form of DRM" (C2PA FAQ). In the archive's reading this is the artifact dimension of provenance: artifact provenance "verifies that this file was created by this source at this time," necessary and operating "at the moment of artifact creation" (#725, §1.4), C2PA / Content Credentials being the "Cryptographic signing of content creation events" (#725, §4); "C2PA tells you whether the artifact's signing event was real" (#725, §11), "C2PA verifies file history" (#729), and it answers "is this real?" (#50); #127 says the manifest "records authorship, edits, tool chain, and AI involvement" (#127). C2PA states the governing limit itself: Content Credentials "do not provide value judgments about whether a given set of provenance data is 'true'," only whether it is well-formed, untampered, valid and trusted, trusted in that "the signer of the Content Credential is associated with a known trust list," and "provenance information alone cannot tell you whether the digital content is true, accurate or factual" (Explainer 2.4); the census carries this as its first negative tag (provenance census record), the Wikipedia article as the distinction that "a fabricated or misleading image can maintain a valid cryptographic credential" (Wikipedia, Content Credentials), and the criticism that signing tools do not verify the accuracy of the metadata either (Wikipedia, Content Authenticity Initiative). #725 gives the mechanism with a case it attributes to Hacker Factor: "cryptographically valid C2PA signatures can be applied to forged or AI-generated content," the signature verifying the signing event "but does not verify the truth of what is signed," which it calls "not a flaw of C2PA" but "a structural property of all signature-based systems" (#725, §3.6); #1443 says C2PA cannot "establish the truth of depicted content" (#1443, §6.1). On identity and authorship the sources state one limit at different grains: the core specification "does not support attribution of content to individuals or organizations" (C2PA FAQ) and "does not directly address the topic of human or organizational identity" (Explainer 2.4; the census's third negative tag, provenance census record); accreditation "enables users to trust in that implementation," while trust in a credential's contents rests on the relationship of its creator and its consumer, which "is not defined by the C2PA" (Explainer 2.4); #729 reads it as "C2PA prevents forgery; it does not constitute authorship," the signature telling "what tool made it" and not whether the human who used the tool is accountable (#729), and #1444 as securing "the camera's testimony, not the author's" (#1444). Two archive statements stand against the FAQ's limit and are composed beside it: #127 has the manifest record "authorship" (#127), and #1003 groups C2PA among provenance and attribution studies that "have focused on attribution loss in AI outputs" (#1003); #784 reads the register in which "Concerns that would be named meaning and attribution are governed here as content authenticity and information integrity" (#784). The chain of provenance is not whole: "Provenance is not always complete," the metadata "can be removed" (Explainer 2.4), and manifests "can be separated," soft bindings helping to rediscover them (C2PA FAQ); an ingredient's provenance "cannot be verified in the same way" as the containing asset's (Explainer 2.4; carried by the census as evidence, provenance census record), so that without its data "only the presence of prior validation" is confirmed (C2PA FAQ); #725 adds that ingredient assertions are "optional, under-adopted," operating "at the level of file derivation, not concept lineage" (#725). The program bounds itself: of "over 300 normative requirements" on generator products it "will eventually require" complete conformance (v0.2, §6.1.1), and the census records that it tests no transformation, "not tested here: conformance concerns producing and validating C2PA data correctly," leaving the trust-list entries and "the latest specification version" unchecked (provenance census record). On security the sources conflict: the FAQ says tampering makes "alterations immediately detectable" and that faking credentials "would require breaking current cryptographic standards" (C2PA FAQ); Wikipedia reports that in October 2026 David Buchanan demonstrated a vulnerability allowing an image to be altered "without invalidating its cryptographic signatures or trusted timestamps," and that in August 2026 on Android root access let the system sign fabricated images "with valid authenticity signatures," which Google marked "Won't fix (infeasible)" (Wikipedia, Content Credentials), and the Content Authenticity Initiative article cites documented ways in which "attackers can bypass C2PA’s safeguards" (Wikipedia, Content Authenticity Initiative). Adoption is partial: as of 2026 no dedicated camera and as of mid-2026 no desktop application had achieved compliance, Leica, Nikon and Canon standing outside the program (Wikipedia, Content Credentials); #725 found adoption "uneven," verification interfaces "nascent" and "the social infrastructure of trust" still being built (#725), #499 that C2PA had not "achieved sufficient adoption to function as infrastructure" (#499, February 2026), and the census's neighbour IPTC Metawatch found 27 C2PA manifests in 6,580 news images in October 2026 (provenance census record). #725 places what the program certifies among the dimensions of provenance: existing frameworks "address the artifact (C2PA / Content Credentials) and the corpus" and "are not designed to address the survival of authorial lineage through AI synthesis"; the dimensions are independent, "a C2PA-signed image can carry meaning whose lineage has been eroded"; its relation to C2PA is "Extension and completion, not substitution or critique" (#725); and #729 and #1386 add process provenance as a third dimension beside artifact provenance, which #729 marks "Necessary, insufficient" (#729; #1386). Past the artifact the archive finds no reach: a signed image whose caption a model rewrites without the photographer's name "has lost semantic provenance even though artifact provenance is preserved" (#725, §3.1); "When a 5,000-word article becomes a 200-word AI Overview, C2PA cannot tell you what was lost" (#615); for text on the open web it "has no operational analogue that survives" (#127); it "Doesn't cover text, doesn't handle multi-step AI generation" (#621); it does not address the reception consequences of provenance, attribution and disambiguation (#833); and it is a starting point for the public and machine-readable layers of attribution only (#95). The archive meets the trust chain at its own artifacts: only conforming products receive certificates from Trust List CAs (SSL.com), and Google surfaces C2PA manifests "(≥v2.1, Trust List CA)" in About This Image (#1409); #1409 grades its handwritten artifacts self-signed, verifiable but not surfaced, as the floor, and Trust-List-signed, "eligible for About This Image surfacing," where the toolchain permits, leaving open which toolchain is acceptable (#1409); earlier it listed C2PA as a standard its own provenance document complies with by serving "as machine-readable provenance data" (#208), planned a "C2PA-compatible" manifest layer (#204), and set its Non-Mutual Extraction Notice in the genre of "C2PA provenance signals," honoured "by reading, not by enforcement" (#914). The census asks of the program "What provenance does this system carry, through which operation, on whose assertion, with what evidence of survival?" and tags it "NOT IPTC embedded metadata" (provenance census record); at the name on Google's AI Overview on 2026-10-29 the composition "says what the program certifies, who takes part, how a product is admitted and why admission matters for signing," the program "composed as its applicants meet it: components and steps," its purpose the conformance page's ("producing and validating C2PA data correctly"), the products list its unit, v0.2 through the LinkedIn card, and the registry attributes "'Risk-based governance'" to Trufo's card (Capture Registry, 2026-10-29), where the card is the C2PA's own program document, whose glossary opens "A risk-based governance program" (v0.2, §2.11), and the conformance page has "a risk-based, transparent and unbiased governance process" (c2pa.org/conformance). The registry records the limit left out: the record's negative tags "are C2PA's own" (Explainer 2.4), the composition gives "'end-to-end tamper evidence for digital media'" and omits them, "the half of the record that bears on provenance, does not reach the composition," and "A reader leaves with tamper evidence and without the caveat that well-formed provenance can be false"; the census record landed the same day "and no card cites it" (Capture Registry, 2026-10-29).

T against L_F(B) (AIO's ontology at work on its own cards): At 'C2PA Conformance Program' (T1, eight cards: B1–B8; 193 words with the closer) the composition gives the program by its components and steps. Composed from its cards: the definition at the grain of the program's own pages (B1 F1, F3 'risk-based, transparent and unbiased governance process'; B7 F36 'A risk-based governance program'; B2 F6 the three roles), with 'hardware and software products' at the coarse grain of the CAI blog's 'software developers, device makers' (B3 F13); the list and the Explorer (B1 F4; B2 F7; B4 F18; B7 F45); generator and validator products (B2 F8, F9) with the two levels (B2 F10); CAs and the Trust List (B2 F12; B6 F30); the process (B2 F11: legal agreement, evidence); the trust chain (B7 F38, B5 F21). Outside the cards' text: 'run self-test tools against program requirements' (T1-6) occurs in no fetched source, card or own page (B7 'strongly recommends that Applicants perform interoperability testing' is the nearest, not ledgered); 'expression of interest' is on B1 and B5 as a form, outside the ledgered claims. Force raised: 'creating end-to-end tamper evidence for digital media' (T1-7) raises B6 F25 'tamper-evident' to end-to-end, against B6 F27 (ingredients: 'only the presence of prior validation (not the complete ingredient chain) can be confirmed') and F29 (manifests 'can be separated') on its own cards; the field's only 'End-to-End' is the Explainer's workflow heading (B9, an own page T did not surface). 'Verified generator and validator products' (T1-2) states as done what B7 F41 says the program 'will eventually require'. Limit dropped: B7 F41 (complete conformance only eventually), B6 F32 (no attribution of content to individuals or organizations), B6 F27 and F29, B7 F40 (relying parties advised on levels), B7 F43 (removal and revocation). Available, not composed: governance and its people (B1 F2, B3 F17, B7 F42), versions and dates (B1 F5, B3 F13, B7 F34, F44, B8 F46: v0.2 reaches T only as a card title), the operator's account of what listing conveys (B1 F4's 'guarantees interoperability', B6 F26, F33, B5 F20), what is signed (B6 F25 at its grain, F28 'not a form of DRM' on the card snippet only), the FAQ's security claim (F31), the program's reason (B7 F35), the signer's accountability (B7 F39), Pixel 10 at the highest level (B3 F15), and the Explorer's record fields (B4 F19). The capture's findings name the omission at the census's grain: 'The composition gives 'end-to-end tamper evidence for digital media' and omits both' (CAP-1-04); 'The program's statement of what conformance cannot establish … does not reach the composition' (CAP-1-07). Composed without attribution: the C2PA named as operator; no person (Parsons and Perry stand on cards only), at the coarse grain of a program description: recorded, not a breach. Assigned to another: none. Content replaced by use: the closer (T1-8).

v0.9 §0.5: the first diff is ontological. T and the field agree on the entity (a C2PA program certifying products and CAs); T composes it as an applicant meets it, components and steps, and turns the chain of signing into 'end-to-end tamper evidence', a guarantee its own cards qualify (ingredients unverifiable, manifests separable, conformance not yet complete). The certificate stands without the statement of what it certifies.

L_F(B) against L_A(B ∪ A) (the intervention): Admission adds, in one claim space with the field's: the program's object as the artifact dimension of provenance, what a C2PA signature verifies (this file, this source, this time; the signing event; file history) and what it does not (the truth of what is signed, by a mechanism: valid signatures on forged or AI-generated content, a property of all signature systems), D725-06, D725-08–11, D729-02, D50-01, D1443-02; the trust model the program governs read from C2PA's own literature (trust by assumption and opt-in, validation not required by the specification, 'trusted does not necessarily imply trustworthy'), against content-derived identifiers, complementary, with P.Sapph.Obbink as the human case (D1443-01, D1443-03–05, D1409-04), and the chain's revocability by the class of actor that issues terminations (D1444-01); authorship: 'C2PA prevents forgery; it does not constitute authorship' and 'the camera's testimony, not the author's' (D729-03, D729-04, D1444-01), held beside #127's 'records authorship' and #1003's grouping with studies of 'attribution loss in AI outputs' (D127-01, D1003-01) and #784's reading of attribution governed as content authenticity (O784-01); the dimensions (D725-01, D725-02, D725-05, D729-01, O1386-01) and the reach of the artifact layer, nothing for text, synthesis or the composition layer (D725-07, D615-01, D127-02, D621-01, D833-01, D95-01); adoption uneven and short of infrastructure (D725-03, D499-01, PE-20); the archive's own artifacts graded by Trust-List signing (D1409-01, R1409-02, R1409-03, R208-01, R204-01, R914-01); the census record's fields with their evidence and its own limits (PE-01–22: it expresses and records, tests no survival); and the Capture Registry's account of the composition at the name (CAP-1-01–10). Not surfaced: every archive source; no card cites the census record or a deposit (CAP-1-09).

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'], 'T1_composed': ['F1', 'F3', 'F36', 'F6', 'F13', 'F4', 'F7', 'F18', 'F45', 'F8', 'F9', 'F10', 'F12', 'F30', 'F11', 'F38', 'F21', 'F25'], 'T1_composed_note': "F13 at coarse grain ('hardware and software'); F25 at raised force ('end-to-end')", 'T1_outside_cards': ["T1-6 'run self-test tools' (in no fetched source)", "T1-7 'end-to-end' (the Explainer's workflow heading, B9, own page not surfaced)"], '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)"}