Laws compared
A cell records what a text requires of the parties it defines. It does not show that any developer falls within them, or complied.
What does each binding government instrument that addresses developers of frontier or general-purpose AI models say on the same questions? Each column is one instrument; each row is one question, worded so that it fits every instrument. The questions are Selekoa’s; an instrument’s own terms appear only in quotation marks. Some instruments address providers of services rather than developers, and the first row quotes the party each one defines.
Question Q01Who is covered
On whom does the instrument place the duties compared here, in its own defined terms?
Criterion: The defined term for the party that bears the duties (developer, provider, company, exporter), quoted from its definition or from the provision that imposes the duties. Where the instrument has tiers, each tier is named in its own defined terms.
What the statuses mean
- Stated
- A passage of the instrument meets the question’s criterion as worded, however conditional or narrow its scope. The condition or scope is in the note.
- Stated, partly
- A passage has what the criterion asks for, but in a weaker form than the criterion’s own words: for example, a power (“may”) where the criterion asks for a duty (“shall”), or a duty without the deadline the question asks about.
- Not found in the recorded search
- No passage meeting the criterion was found. The cell lists the sections read, the search terms and every passage set aside, with the reason. It does not show that the instrument lacks such a provision elsewhere, or that no other law imposes one.
- Does not apply
- The question does not arise for this instrument, for the reason given (for example, there is no framework document whose publication could be asked about).
Four rules decide “partly” in borderline cases:
- An order to an agency to impose a duty counts as partly.
- Protection only by reference to another law counts as partly.
- “Without undue delay” counts as a stated deadline.
- Required security measures for model weights count as risk-management measures under Q03.
Every stated or partly cell quotes the instrument’s own text with its article, section or page; Chinese passages carry Selekoa’s working translation. Three cells give a paraphrase with its locator instead, and say so. A not-found cell lists the sections read, the terms searched and every passage set aside, with the reason.
Paraphrased with its locator, not quoted: quoting it would pass the source’s quotation limit, or the source’s length is not known.
Limits of this comparison
- A cell records what the text requires of the parties it defines. It does not show that any named developer falls within them, or that anyone complied.
- A not-found cell is bounded by its search: the sections read and the terms searched. Another provision, or another law, may still impose the duty.
- The Code of Practice is a compliance route, not a law. Executive Order 14110 is revoked, and the BIS rule’s rescission was announced; each column shows its standing.
- The table compares and does not rank. Statuses are not scores and are not to be summed.
- Biological and chemical risk provisions are recorded at policy level only.
- Dates keep their kind: adoption, signature, approval, publication, entry into force and application are different events. A computed date is marked as computed.
How we searched
The instruments started from those on the frameworks-and-law page. A column was kept where the instrument is binding, or is the AI Act’s stated means of compliance (the Code), and places duties on developers or providers of frontier or general-purpose AI models. Fifty other instruments on that page were set aside, each with its reason, in the research pack: proposed or vetoed texts, guidance and reports, voluntary commitments, binding texts that place no duty on developers, and one standard whose text was not retrieved (GB 45438-2025).
United Kingdom: within the searches made, no UK law or other binding instrument placing duties on developers of such models was located. This is not a finding that none exists. The 2024 King’s Speech announced an intent to legislate; the 2026 briefing notes contain none of the searched terms; the memoranda with developers state that they are not legally binding; and title searches of legislation.gov.uk for 2023 to 2026 found one instrument, which places a duty on the Information Commissioner.
Other US federal texts: as read, none places a binding duty on developers. Executive Order 14409 of 2 June 2026 provides for a voluntary access framework and says that its section 3 authorises no licensing requirement.
Not yet read: the Republic of Korea’s AI Basic Act, Japan’s AI Promotion Act (2025), China’s provisions on deep synthesis (2022) and on algorithm recommendation (2021), and Colorado SB 24-205 (2024). They are not in the dataset; a later pack starts with Korea.
Illinois: the act’s official website refused the preparer’s requests, and the Internet Archive reset the connections. The cross-family checker’s session saved the enacted text from an Internet Archive capture of the official PDF (2026-09-08), and the cells were drafted from that saved text. The two model families read the act independently and agreed on the status of 13 of 15 cells. The approval date rests on the checker’s copy of the official bill status record, which the preparer could not open; it is given as a paraphrase, labelled as that copy.
Texts were read in official copies: the EU texts through the Publications Office’s repository, because EUR-Lex returned empty responses; California SB 53 from the Legislative Counsel’s database file, because its website refused requests; New York’s chapters in the Senate print and the Assembly’s bill records, not the session laws; the Chinese measures in the gov.cn policy library. Six web pages served changed bytes between retrievals; every quotation still matched.
Roles: one model family (Claude) drafted the cells, audited 50 of them in a fresh context and adjudicated. A model from another family (GPT, through Codex) checked 122 items, answering 85 of them blind before seeing the drafted answers. Where it disagreed, its status was applied. One cell it disputed (Executive Order 14110, Q06) was settled as not found after its second pass. For Illinois the roles were reversed: the checker read the act first, blind, and the drafter then wrote all fifteen cells without seeing its answers. The maintainer decided the remaining questions, including the four rules for “partly” and Illinois Q05 (partly, as for California’s identical wording). No person has checked the cells against the texts.
European Union
EU AI Act
Stated
“providers of general-purpose AI models”
Article 53(1), chapeau (same words as in Article 101(1), dataset quote aia-art101-gpai-fines)
“shall be classified as a general-purpose AI model with systemic risk if it meets any of the following conditions”
Article 51(1), chapeau
Note: The duties in Article 53 fall on providers of general-purpose AI models; Article 55(1) adds duties for providers of general-purpose AI models with systemic risk, the category defined in Article 51 (high-impact capabilities, presumed above a training-compute threshold, or a Commission decision). Article 54 adds an authorised-representative duty for providers established in third countries; Article 53(2) exempts certain open-source models from points (a) and (b), but not models with systemic risk. 'Provider' is defined in Article 3(3) (not quoted).
Inspect sources and checks: EU AI Act, Q01 Who is coveredEU Code of Practice, Safety and Security
Stated
“this Safety and Security Chapter (“Chapter”) is only relevant for providers of general-purpose AI models with systemic risk and not AI systems”
Recitals, (b) (PDF p. 3)
Note: The commitments are written as commitments of 'Signatories'; the Chapter is stated to be relevant only to providers of general-purpose AI models with systemic risk. The commitments are addressed to providers that sign the Code and choose to rely on it; the Code is not binding in itself (see the row’s standing). Terms defined in Article 3 of the AI Act apply (Glossary, opening paragraph).
Inspect sources and checks: EU Code of Practice, Safety and Security, Q01 Who is covered
United States: states
California SB 53
Stated
“together with its affiliates collectively had annual gross revenues in excess of five hundred million dollars ($500,000,000) in the preceding calendar year.”
SEC. 2, Bus. & Prof. Code § 22757.11(j)
“has trained, or initiated the training of, a frontier model”
SEC. 2, Bus. & Prof. Code § 22757.11(h)
Note: Two defined terms bear the duties: “frontier developer” (§ 22757.11(h)) and “large frontier developer” (§ 22757.11(j)), a frontier developer whose revenues with affiliates exceeded $500 million in the preceding calendar year. The framework duty (§ 22757.12(a)), the transparency-report summaries of risk assessments (§ 22757.12(c)(2)), internal-use summaries (§ 22757.12(d)) and the anonymous internal process (Lab. Code § 1107.1(e)) fall on large frontier developers; the transparency report (§ 22757.12(c)(1)), incident reporting (§ 22757.13(c)) and the anti-retaliation rule (Lab. Code § 1107.1(a)) fall on all frontier developers.
Inspect sources and checks: California SB 53, Q01 Who is coveredNew York RAISE Act (As amended, Chapter 96 of 2026)
Stated
"Large frontier developer" means a frontier developer that … collectively had annual gross revenues in excess … of five hundred million dollars in the preceding calendar year”
Bill text, § 2 (GBL § 1420(10) as re-added), p. 3, lines 53–55
"Frontier developer" means a person who has trained, or initiated”
Bill text, § 2 (GBL § 1420(8) as re-added), p. 3, line 41
Note: Two defined terms bear the duties: “frontier developer” (transparency report, incident reporting, statements) and “large frontier developer” (framework, internal-use summaries, disclosure statement under § 1428, penalties under § 1427). § 1426 excludes accredited colleges and universities engaged in academic research and the Empire AI consortium. Ellipses in quotes from this record skip omitted words or the printed line numbers. These duties are in Article 44-B as re-added, which takes effect on 1 January 2027 (Q11).
Inspect sources and checks: New York RAISE Act (As amended, Chapter 96 of 2026), Q01 Who is coveredNew York RAISE Act (As signed, Chapter 699 of 2025)
Stated
"Large developer" means a person that has trained at least one frontier model and has spent over one hundred million dollars in compute costs in aggregate in training frontier models.”
Proposed GBL § 1420(9), p. 3, lines 4–6
Note: One tier: “large developer”. Accredited colleges and universities engaged in academic research are excluded, and a transferee of full intellectual property rights in a frontier model becomes the large developer (§ 1420(9)). Text as signed in Chapter 699; Chapter 96 of 2026 repeals and re-adds Article 44-B (see the amended text).
Inspect sources and checks: New York RAISE Act (As signed, Chapter 699 of 2025), Q01 Who is coveredIllinois AI Safety Measures Act
Stated
“Frontier developer” means a person who trains, or … initiates the training of, a frontier model”
Section 5, definition of “frontier developer”, PDF pp. 3–4
“Large frontier developer” means a frontier developer that, together with its affiliates, collectively had annual gross revenues in excess of $500,000,000 in the preceding calendar year.”
Section 5, definition of “large frontier developer”, PDF p. 4
Note: Two tiers in the act's defined terms. The transparency report (Section 10(c)(1)), incident reporting (Section 15(c)), the statements rule (Section 10(f)(1)) and the whistleblower duties (Section 20(a)–(d)) fall on frontier developers; the framework (Section 10(a)–(b)), the risk summaries in the transparency report (Section 10(c)(2)), the audit (Section 10(d)), internal-use summaries (Section 10(e)), the internal reporting process (Section 20(e)) and the disclosure statement (Section 18) fall on large frontier developers. “Person” is defined as a nongovernmental individual or organisation (Section 5). “…” in the quotation stands for the running footer at a PDF page break; no words of the act are omitted.
Inspect sources and checks: Illinois AI Safety Measures Act, Q01 Who is covered
China
These measures address providers of services, not developers as such.
China, generative AI services (2023)
Stated
“生成式人工智能服务提供者（以下称提供者）”
Working translation: providers of generative AI services (hereinafter 'providers')
Article 7 (第七条), chapeau
“包括通过提供可编程接口等方式提供生成式人工智能服务”
Working translation: including providing generative AI services by means such as providing programmable interfaces
Article 22 (第二十二条), point (2)
Note: The duties fall on 生成式人工智能服务提供者 (providers of generative AI services), defined in Article 22(2) as organisations and individuals that use generative AI technology to provide generative AI services, including through programmable interfaces. The Measures do not define a developer; they have no tiers. Some provisions (Article 4) address both provision and use; Article 22(3) defines users (使用者). Translations are Selekoa's working translations.
Inspect sources and checks: China, generative AI services (2023), Q01 Who is coveredChina, AI content labelling (2025)
Stated
“符合…规定情形的网络信息服务提供者（以下简称“服务提供者”）”
Working translation: network information service providers that fall within the circumstances provided for in [the three named rules] (hereinafter 'service providers')
Article 2 (第二条)
“开展人工智能生成合成内容标识活动，适用本办法”
Working translation: [when they] carry out labelling of AI-generated synthetic content, these Measures apply
Article 2 (第二条), closing words
Note: The principal duties fall on 服务提供者 (service providers): network information service providers within the circumstances of the Algorithmic Recommendation Provisions, the Deep Synthesis Provisions or the Interim Measures for Generative AI Services. Other duties fall on service providers offering content-dissemination services (Article 6), internet application distribution platforms (Article 7), and users and 'any organisation or individual' (Article 10). The Measures do not address model developers as such and have no tiers.
Inspect sources and checks: China, AI content labelling (2025), Q01 Who is covered
Revoked, or rescission announced
Kept as part of the record, each with its standing shown.
US Executive Order 14110
Stated, partly
“the Secretary of Commerce shall require:”
Sec. 4.2(a), opening words, 88 FR 75197
“Companies developing or demonstrating an intent to develop potential dual-use foundation models to provide the Federal Government, on an ongoing basis, with information, reports, or records regarding the following”
Sec. 4.2(a)(i), 88 FR 75197
Note: The order does not itself place duties on companies; it directs the Secretary of Commerce, acting under the Defense Production Act, to require “Companies developing or demonstrating an intent to develop potential dual-use foundation models” to report. The term “dual-use foundation model” is defined in Sec. 3(k); Sec. 4.2(a)(ii) separately covers those who acquire, develop or possess a potential large-scale computing cluster. Search for the rest of the criterion: Whole cached text of the Federal Register printing (88 FR 75191–75226), with close reading of Secs. 3, 4.1, 4.2, 12 and 13 and the signature block. Terms searched: “shall require”, “Companies”, “dual-use foundation model”. Passages read and not counted: Sec. 4.2(c), 88 FR 75197–75198: Directs proposed regulations on United States IaaS Providers concerning foreign persons' training runs; not the developer reporting compared here.
Inspect sources and checks: US Executive Order 14110, Q01 Who is coveredUS BIS AI diffusion rule
Stated
“exporters, reexporters, and transferors”
DATES, 90 FR 4544
“The VEU may only store or transfer the model weights”
Supplement No. 10 to part 748, para. 8, 90 FR 4565 (PDF p. 22)
Note: An export-control rule: duties fall on exporters, reexporters and transferors of controlled items, including model weights specified in ECCN 4E091 (licence requirement in § 742.6(a)(13); conditions of License Exception AIA in § 740.27), and on Data Center Validated End Users (VEUs) under Supplement No. 10 to part 748. The rule has no category of model developer as such.
Inspect sources and checks: US BIS AI diffusion rule, Q01 Who is covered