What the New York RAISE Act Requires of AI Developers

What the New York RAISE Act Is and What Starts in November and January

The New York RAISE Act is a state law that sets safety rules for companies building frontier artificial intelligence (AI) models. RAISE stands for Responsible AI Safety and Education. Large frontier AI developers must register with a new state office, publish safety frameworks and report critical safety incidents within 72 hours.

That office is the Office of Digital Innovation, Governance, Integrity and Trust (DIGIT), and it sits inside the New York State Department of Financial Services (DFS). In her September 21, 2026 announcement on AI safety, Governor Kathy Hochul set out the next two dates.

"Starting in November, New York will direct large frontier AI developers to register with the State."

"Beginning January 2027, AI companies will be required to comply with those strict standards."

The law reached this point in five steps.

1. December 19, 2025: Hochul signed the initial RAISE Act, with her approval conditioned on the Assembly passing chapter amendments.
2. March 27, 2026: Hochul signed the final version, which includes those chapter amendments.
3. September 21, 2026: Hochul announced next steps, including the DIGIT office and its deputy director.
4. November 2026: Large frontier AI developers begin registering with the state, according to the Governor.
5. January 1, 2027: The law takes effect, according to a summary by the law firm Davis Polk.

The November step is registration, and the Governor's page ties compliance to January 2027.

Who Has to Register Under the RAISE Act

The RAISE Act's heaviest duties fall on large frontier developers, a phrase the Governor's page uses without defining. Her announcement gives no revenue or computing thresholds. The definitions below come from the Davis Polk summary of the final RAISE Act.

In Davis Polk's description, a frontier model is one trained with more than 10^26 integer or floating-point operations. Computing power from the original training and from later fine-tuning both count toward the total.

A large frontier developer is one with annual gross revenues above $500 million in the last calendar year. Davis Polk says this threshold matches California's law. Some duties apply to every frontier developer, and others apply only to large frontier developers.

None of the sources cited here name the companies that must register.

What Developers Must Do Under the New York RAISE Act

The New York RAISE Act splits its duties into two tiers, according to Davis Polk. Every frontier developer carries a short list, and large frontier developers carry a longer one.

Duties for All Frontier Developers

• Publish a transparency report when deploying a new or substantially modified model.
• Make no materially false or misleading statements about catastrophic risk.
• Report critical safety incidents to DFS within 72 hours, or within 24 hours if there is imminent risk of death or serious injury.

Extra Duties for Large Frontier Developers

• Write, implement and publish a frontier AI framework, review it every year, and republish it when changes are material.
• Make no materially false statements about following that framework.
• File a disclosure statement with DFS before developing, deploying or operating a frontier model, and renew it every two years or after material changes.
• Pay a pro rata share of the costs of administering the Act.
• Send DFS summaries of catastrophic-risk assessments for internal use every quarter, unless DFS sets another schedule.

The Pillsbury summary of the RAISE Act also lists the duties. Its list starts with registering with DIGIT and publishing a safety and transparency framework on the developer's website. Pillsbury also says the disclosure statement is due at least once every two years starting January 1, 2027.

The Governor's announcement describes the duties more briefly: companies must publish their safety protocols and report regularly to DIGIT. None of the three sources gives a dollar amount for the assessments developers pay toward the state's oversight program.

What DIGIT Is and How It Fits Inside DFS

DIGIT is the office inside DFS that administers the RAISE Act. According to Davis Polk, DIGIT receives disclosure statements, reviews risk assessments and incident reports, and reports each year to the Governor and the Legislature. DIGIT also has notice-and-comment rulemaking authority, which means it can write rules under the Act after public notice.

Kaitlin Asrow is Acting Superintendent of DFS. Marc Gilman is Deputy Director of DIGIT. Governor Hochul says Gilman has years of experience in technology risk management and compliance, and previously served as general counsel overseeing regulatory issues for financial firms.

Members of the public can report suspected critical safety incidents to DIGIT, Pillsbury reports. DIGIT must also publish an annual report summarizing the incident reports it receives and its own observations. For AI incidents documented so far, see the Ban the Bots AI incident tracker.

Penalties and Enforcement Under the RAISE Act

The Governor's page says "Companies that don't play by the rules will face civil penalties," without giving amounts. Davis Polk's summary of the final law supplies the figures.

• The New York Attorney General can seek civil penalties against large frontier developers of up to $1 million for a first violation.
• The Attorney General can seek up to $3 million for subsequent violations.
• DFS can impose $1,000 per day of noncompliance for failing to keep disclosures current or pay assessments, plus the assessments owed.

The Act expressly disclaims a private right of action. A person who suspects a critical safety incident can report it to DIGIT, according to Pillsbury.

How the RAISE Act Compares with California's Law

New York's chapter amendments were agreed to bring the RAISE Act more in line with California's Transparency in Frontier Artificial Intelligence Act, according to Davis Polk. Pillsbury names California SB 53 (Senate Bill 53), enacted in 2025, as the comparable law. Davis Polk lists four ways New York's version differs from California's:

• Incident reporting: New York allows 72 hours to report a critical safety incident, compared with 15 days in California.
• Rulemaking: New York gives a state office the authority to write rules under the Act.
• Whistleblowers: The RAISE Act has no whistleblower protections.
• Disclosure statements: New York requires large frontier developers to file a disclosure statement with DFS.

For how state AI laws sit alongside national rules, read the Ban the Bots explainer on state versus federal AI regulation.

What to Watch Before January 2027

Four parts of the RAISE Act are open or still ahead, based on the sources cited here.

• DIGIT rules: DIGIT can write rules through a notice-and-comment process, and none of the sources describe rules it has issued.
• November registration: The Governor has said large frontier AI developers start registering in November, with compliance beginning January 2027.
• Public thresholds: The Governor's page gives no definition of large frontier AI developers, so the $500 million and 10^26 figures come from Davis Polk's description of the final law.
• The annual report: DIGIT must publish a yearly summary of incident reports and its observations, according to Pillsbury.

Where to Read More About the New York RAISE Act

Pillsbury names California SB 53 as the law comparable to the New York RAISE Act. Read the Ban the Bots guide to AI regulation for the wider picture, then compare New York's approach with the Colorado AI law explainer.

This page provides general information about the RAISE Act and does not give legal advice.