New York opens RAISE Act portal for frontier AI labs

By Grace Kim

What does the Jan. 1 deadline require?

New York opened its RAISE Act registration portal on Tuesday, according to TechCrunch, setting a Jan. 1, 2027 deadline for large frontier AI developers such as OpenAI and Anthropic. The deadline coincides with the law's effective date. The state can take enforcement action against companies that fail to register, although a spokesperson declined to specify the available actions.

Registration goes to the Office of Digital Innovation, Governance, Integrity and Trust, or DIGIT, a new unit within the Department of Financial Services. DIGIT will oversee the labs.

Gov. Kathy Hochul said in September that registration would begin in November. She named Marc Gilman, formerly general counsel at Theta Lake, as DIGIT's deputy director for the RAISE Act and its first full-time hire.

What must covered labs disclose?

The law applies to developers that train models above 10^26 computational operations and have more than $500 million in annual revenue, two thresholds that together define the covered class. Both conditions must be met.

• Publish safety and transparency frameworks on their websites.
• Report critical safety incidents to DIGIT within 72 hours.
• Submit quarterly assessments of catastrophic risks.
• File a disclosure statement at least every two years, beginning Jan. 1, 2027.
• Pay assessments that fund DIGIT.

Fines can reach $1 million for a first violation and $3 million for subsequent violations.

Members of the public can also report suspected incidents to DIGIT. The office will publish an annual report on the reports it receives.

The statute uses training compute and revenue to define coverage. The announcement did not identify every covered lab. It also gave no assessment amount or company-specific compliance estimate. Those financial figures remain undisclosed.

What will test advanced systems?

The state also launched the New York Center for AI Responsibility and Research at Binghamton University to independently evaluate advanced AI systems and assist DIGIT's reviews of developer safety frameworks. MLQ.ai covered that testing program separately. It is a state research effort, distinct from the RAISE Act's registration and reporting duties.

That separation matters because statutory filings and independent model evaluations create different channels for scrutiny. The filing track sets recurring deadlines for labs. The research center supplies technical evaluations that can support DIGIT's framework reviews. Neither track replaces the other.

How does New York compare with other states?

California's SB 53 took effect in January and covers the same kind of large frontier developer, with 15 days to report a critical incident and penalties capped at $1 million per violation. New York's 72-hour window falls to 24 hours when there is imminent danger of death or serious injury. DIGIT also has direct rulemaking power.

Illinois's SB 315 takes effect in 2027 and adds independent safety audits. Outside the United States, Australia said this week that it will create an AI safety regulator for frontier labs. OpenAI and Anthropic signed France's Paris Call on fail-safe shutdowns.

The comparison creates a concrete timing difference: New York requires initial notice within three days, while California allows 15. The 24-hour New York rule applies when there is imminent danger of death or serious injury. Illinois adds another procedural layer through independent safety audits in 2027.

Australia's planned regulator and the Paris Call show that frontier-model governance now spans state, national and international commitments. Companies subject to both regimes would need separate response procedures to meet both clocks.

By Jan. 1, 2027, covered labs will face a shared competitive requirement: registration, recurring disclosures and rapid incident reporting under New York's enforcement regime.