New York opens RAISE Act sign-ups for frontier AI labs

New York just started the clock on frontier AI labs.

The state opened registration on Tuesday for the RAISE Act, its AI safety law, and launched a public research center to test advanced AI systems. Large frontier developers have until Jan. 1, 2027 to register.

What New York launched

New York's Department of Financial Services announced the RAISE Portal, an online registration system run by its Office of Digital Innovation, Governance, Integrity, and Trust, known as DIGIT.

Large frontier AI developers must register through it by Jan. 1, 2027.

The same release launched the New York Center for AI Responsibility and Research at SUNY Binghamton. DFS says it will test advanced AI systems, do foundational research and train AI safety experts, with more than 80 researchers across the state and support from Empire AI. Greg Morrisett and Jeannette Wing are interim co-directors.

"The stakes are too high to simply take AI companies at their word," Gov. Kathy Hochul said in the release.

What the law asks of the labs

DFS says large frontier developers must publish safety frameworks on their websites beginning in January. All frontier developers must report critical safety incidents and publish model transparency reports.

The reporting deadline is short.

72 hours from a critical safety incident.

Hochul's office said in September that developers must also file quarterly assessments of catastrophic risks with DIGIT, that the public may file suspected incident reports, and that DIGIT will publish an annual public report.

TechCrunch reports that the law covers companies with more than $500 million in annual revenue, names Anthropic and OpenAI among them, and puts penalties at up to $1 million for a first violation and up to $3 million for later ones. TechCrunch also notes California's SB 53 gives companies 15 days to report a critical incident.

Why a 72-hour rule matters to you

Most of this is paperwork for the labs. The clock is the part with teeth.

If a model you depend on does something dangerous in deployment, the lab now owes New York a report within three days, and the state publishes what it learns each year. That is faster than California's rule, and faster than most companies tell their own customers anything.

Be clear about the limits. Engadget notes the new center can only study what AI companies disclose publicly.

New York is also not alone this week. Australia proposed its own AI safety regulator for frontier labs, as TopFive reported yesterday.

What to watch

Watch the registry after Jan. 1. Which labs sign up, and whether any large developer argues it is out of scope, will show how far the law reaches.

Then watch for DIGIT's first annual report. It is the first public look at what the labs had to confess.

Jan. 1 is the deadline. The first incident report is the test.