New York signs the RAISE Act to regulate AI safety

When the federal government stalls on tech regulation, state governors tend to grab the steering wheel. New York has officially stepped into the arena.

Governor Kathy Hochul just signed the RAISE Act (Responsible AI Safety and Education Act), making New York the second U.S. state behind California to enact major frontier AI safety rules.

The core mandate requires developers of frontier AI models to publish safety protocols and report critical safety incidents within 72 hours. The law also sets up a dedicated oversight office inside the Department of Financial Services. Skirting the disclosure rules or filing false statements can trigger fines up to $1 million for a first offense and $3 million for repeat violations.

A bruised bill crosses the finish line

Getting the ink on the paper involved a massive tug-of-war. New York lawmakers initially passed the bill back in June, but aggressive tech lobbying immediately pushed Hochul to pump the brakes.

The AI Alliance, a coalition that includes Meta, IBM, Intel, and several top universities like Cornell and NYU, spent tens of thousands of dollars on digital ads warning that the law would stifle 400,000 tech jobs. At the same time, more than 150 parents sent letters urging Hochul to sign the original text.

The resulting legislation came out significantly softer. The Verge reports that lawmakers dropped a crucial clause that would have banned frontier models carrying unreasonable risks of catastrophic harm, specifically scenarios causing 100 or more deaths or over $1 billion in damages from biological or nuclear weapons. The New York Times reported that Hochul signed the original text after lawmakers pledged to enact rollbacks next legislative session.

Public praise and behind-the-scenes PACs

The reaction across Silicon Valley has been split right down the middle.

OpenAI and Anthropic publicly backed the bill while continuing to plead for uniform federal standards. Anthropic head of external affairs Sarah Heck noted that having the country’s two biggest economic hubs pass transparency rules sends an undeniable message to Capitol Hill.

Yet the backchannel fight tells a far more aggressive story. A super PAC backed by Andreessen Horowitz and OpenAI president Greg Brockman is actively preparing to challenge Assemblyman Alex Bores, the bill co-sponsor, in his upcoming primary.

All of this sits on a collision course with Washington. President Donald Trump recently issued an executive order directing federal agencies to invalidate state-level AI mandates, a push championed by White House AI czar David Sacks that will almost certainly land in federal court.

Why it matters

For regular developers and enterprise buyers, this law targets the handful of players training foundational models rather than smaller apps downstream. But mandatory 72-hour incident reporting means safety failures will soon have to be aired out in public rather than swept under corporate rugs.

We think the frantic lobbying campaign highlights the fragile nature of state-level AI oversight. Stripping out the severe liability provisions around catastrophic harm was clearly the political price New York had to pay to get anything enacted. It makes the final text look much softer than state sponsors originally bragged, but establishing an official government monitoring office is still a meaningful foot in the door.

The immediate headwind is regulatory fragmentation. With California and New York establishing separate guardrails while the White House prepares preemption lawsuits, AI builders will soon find themselves navigating a messy patchwork of conflicting state and federal demands.