FRONTIER · ARTICLE 44-B

New York’s Article 44-B,
implemented in the open.

An independent reference implementation of the RAISE Act’s safety and transparency regime for frontier AI developers — the published frameworks, the statutory text, and an eligibility check, all in one place. The regime takes effect January 1, 2027.

FilingsGPT is an independent reference implementation of New York's Article 44-B (the RAISE Act). It is not affiliated with or operated by the NY Department of Financial Services or the State of New York, and is not an official filing system.

44-B
Article of the NY General Business Law
10
Statutory sections (§§ 1420–1429)
6
Published safety frameworks tracked
Jan 1 2027
Effective date of the regime
RegimeNY Article 44-BEffectiveJan 1, 2027OversightNY DFSEnforcementNY Attorney GeneralIncident reporting72h / 24hRecord retentionDeployment + 5 yrsFrameworks published6Statutory sections§§ 1420–1429RegimeNY Article 44-BEffectiveJan 1, 2027OversightNY DFSEnforcementNY Attorney GeneralIncident reporting72h / 24hRecord retentionDeployment + 5 yrsFrameworks published6Statutory sections§§ 1420–1429

Who the Act covers

Start with one question.

Published frameworks tracked
0

Article 44-B takes effect January 1, 2027. If you build frontier AI models, the first question is simple — are you a covered developer?

Likely covered

Developers that train a frontier model — a foundation model trained with more than 10²⁶ operations — and that, with affiliates, exceed $500 million in annual revenue (a 'large frontier developer'), where the model reaches New York.

Generally outside

Smaller developers below the revenue bar carry only lighter duties; academic research at New York institutions and the Empire AI consortium are excluded entirely. Thresholds are from the enacted text; DFS rulemaking fills in operational detail.

Disclosure · Frontier AI Framework (redacted)Sample
Filed byIllustrative large frontier developer
Framework elementDisclosed
Capability thresholds
Defined, with named eval suites
Risk assessment
Pre-deployment testing on record
Catastrophic-risk mitigations
Deployment & access controls
Incident response
72h reporting path to DFS Office
Governance
Named accountable owner; version history
Update cadence
Reviewed periodically; published redacted
Unredacted framework retained
With testing records · auditable by DFS
Deployment
+ 5 years

What the Act requires

Six duties, one regime.

Article 44-B asks large frontier developers to write down how they manage catastrophic risk, to publish it, to report when something goes wrong, and to register with an oversight office inside NY DFS. The text below tracks the enacted statute (S.8828).

Read all of Article 44-B
§ 1421

Frontier AI framework

A large frontier developer must write, implement, comply with, and conspicuously publish a frontier AI framework describing how it manages, assesses, and mitigates catastrophic risk — reviewed at least annually, with material changes published within thirty days.

§ 1421(3)

Transparency reports

Before deploying a new or substantially modified frontier model, a frontier developer publishes a transparency report — release date, intended use, and catastrophic-risk assessment summaries. A system card or model card satisfies this.

§ 1422

Incident reporting

A critical safety incident is reported to the DFS Office within 72 hours — and within 24 hours to law enforcement if it poses an imminent risk of death or serious injury. Large frontier developers also report internal-use risk assessments every three months.

§ 1420(3)

Catastrophic-risk definition

The bar the regime is built around: a foreseeable, material risk that a frontier model materially contributes to more than fifty deaths or serious injuries, or more than $1 billion in property damage, from a single incident.

§ 1428

Disclosure statement

A large frontier developer files a disclosure statement with the DFS Office — identity, New York addresses, 5%+ owners — renewed every two years. The Office maintains and publishes the list of developers who have filed.

§ 1427

Enforcement

The New York Attorney General enforces the article with civil penalties up to $1 million for a first violation and up to $3 million for subsequent violations. There is no private right of action.

For the record

Questions about the Act.

The RAISE Act is the New York law that adds Article 44-B to the General Business Law (§§ 1420–1429). It creates a safety and transparency regime for developers of frontier AI models — the duty to write, publish, and stand behind a frontier AI framework, to publish a transparency report before deploying, to report critical safety incidents, and to submit to oversight. It takes effect January 1, 2027.

The heaviest duties fall on a 'large frontier developer' — a developer that trains a frontier model (a foundation model trained with more than 10²⁶ integer or floating-point operations) and that, with its affiliates, had annual gross revenue over $500 million in the preceding year. The well-resourced labs training the most capable models. Smaller developers and academic work generally fall outside the heaviest duties. The thresholds are set by the enacted text and DFS rulemaking fills in operational detail, so verify before relying on it.

A written, implemented, and published frontier AI framework describing how the developer manages, assesses, and mitigates catastrophic risk, plus a transparency report before deploying a new or substantially modified frontier model (a system or model card satisfies it). A large frontier developer also files a disclosure statement with the DFS Office, renewed every two years. Where a published document is redacted to protect trade secrets, cybersecurity, public safety, or national security, the unredacted version is retained for five years. Critical safety incidents are reported to the DFS Office within 72 hours — and within 24 hours to law enforcement where the risk of death or serious injury is imminent.

January 1, 2027 (set in § 3 of the act). The final version was enacted by the S.8828 chapter amendment, signed March 27, 2026, and DFS rulemaking will finalize the filing format — so treat operational specifics as provisional until you check them against the enacted text.

No. FilingsGPT is an independent reference implementation of the Article 44-B regime — it is not affiliated with, endorsed by, or operated by the NY Department of Financial Services or the State of New York. There is no official state filing portal yet; this models what one should be.

There is no private right of action. The New York Attorney General enforces the Act through civil penalties — up to $1 million for a first violation and up to $3 million for a subsequent one (§ 1427). Separately, the DFS Office may levy $1,000 per day for a missing disclosure statement (§ 1428). The Office monitors compliance and publishes the list of large frontier developers who have filed.

See where you stand.

Walk the eligibility check, read the statutory sections, or ask the assistant a question grounded in the Act and the six published frameworks. No account required.

FilingsGPT.

Article 44-B · reference implementation

FilingsGPT models a public Article 44-B filing and disclosure system; it does not provide legal advice. Figures, deadlines, and the compute threshold are drawn from the enacted text (the S.8828 chapter amendment that added Article 44-B). The six published frameworks are voluntarily-released lab documents shown as illustrative reference, not actual 44-B filings.

FilingsGPT is an independent reference implementation of New York's Article 44-B (the RAISE Act). It is not affiliated with or operated by the NY Department of Financial Services or the State of New York, and is not an official filing system.