New York starts enforcing its AI safety law, and the biggest labs must register by January 1

New York is getting ready to enforce one of the most significant state AI safety laws in the US. On Tuesday the state opened a registration portal for large frontier AI developers, companies like OpenAI and Anthropic, which must sign up by January 1, 2027.

What the RAISE Act requires

The RAISE Act, New York's AI safety and transparency law, takes effect at the start of 2027. It applies to "large frontier developers," defined as companies with more than $500 million in annual revenue. They will have to:

• publish their safety frameworks on their websites,
• report critical safety incidents within 72 hours, and
• issue public transparency reports about their models.

A newly created state office, the Office of Digital Innovation, Governance, Integrity, and Trust (DIGIT), will carry out the law and oversee the big labs. Penalties can reach $1 million for a first violation and $3 million for later ones. The state can also take action against companies that don't register, though a spokesperson declined to say how.

A new research center

Governor Kathy Hochul also launched the New York Center for AI Responsibility and Research at SUNY's Binghamton University. The center will independently evaluate the safety and security of advanced AI systems and work with DIGIT to review companies' safety frameworks. Hochul first announced it in January, backed by $55 million in donations and capital.

At a press briefing, Hochul framed the law as pushing back on tech executives who present their vision of AI as inevitable, and said New York wants to be a national model for regulation that puts communities ahead of corporations.

Part of a growing patchwork

New York already enforces a law requiring AI companion apps to regularly remind users they're talking to a bot and to make reasonable efforts to detect suicidal thoughts and self-harm. In October, Hochul signed a separate chatbot safety law for minors that restricts things like deceptive human-like behavior and manipulative or sycophantic responses. That one takes effect June 1, 2027.

Other states are moving in the same direction:

Why it matters

With no comprehensive federal AI law, states are setting the rules that the biggest AI companies actually have to follow. New York's 72-hour reporting window is notably tighter than California's 15 days, and the registration requirement gives the state a clear list of who is covered. For the labs, that means another set of filings and deadlines, and a new regulator that can ask hard questions about how their models are tested before release.