New York opens its RAISE Act portal: frontier AI labs must register by January 1, 2027

New York has started the clock on one of the most consequential state AI laws in the country. On November 10, 2026, the state opened a registration portal for large frontier AI developers, with a registration deadline of January 1, 2027. The law behind it, the RAISE Act, takes effect at the start of the year. For OpenAI, Anthropic and any other lab above the revenue threshold, this turns a statute into paperwork with penalties attached.

Who has to register and what they owe

According to TechCrunch, the RAISE Act applies to "large frontier developers," defined as companies with more than $500 million in annual revenue. Three duties sit at the center of the law. These developers must publish safety frameworks on their websites. They must report critical safety incidents within 72 hours. And they must issue public model transparency reports.

Those are reporting and disclosure duties, not a licensing regime. Nothing in the law as described by TechCrunch requires state approval before a model ships, which fits the federal line set out in Executive Order 14409 against mandatory pre-release licensing. What New York has built is a transparency-and-accountability layer: say what your safety plan is, tell us fast when something goes badly wrong, and publish what your models can do.

The enforcer and the penalties

The newly formed Office of Digital Innovation, Governance, Integrity, and Trust, known as DIGIT, will implement the act and oversee the large labs. Penalties can reach $1 million for one violation and $3 million for subsequent violations. The state can also bring enforcement actions against companies that fail to register, although a spokesperson declined to give TechCrunch specifics on how.

That gap is worth noting. A deadline, a portal and a fine ceiling are concrete. How DIGIT will decide that a safety framework is adequate, what counts as a "critical" incident, and whether it will publish what it receives are the details that decide whether the law has teeth. None have been spelled out publicly in the coverage we reviewed.

A state testing center joins the law

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

The governor's own pressroom lists the launch under the headline "Leading the Nation in AI Science and Safety" and calls it a first-of-its-kind statewide effort. One limit to keep in mind, flagged in a Neuron roundup of the day's news: access to unreleased systems is limited. An evaluator that cannot test models before release is judging the safety case a lab hands it rather than the model itself.

Hochul framed the stakes in political terms. "Tech moguls tell us that their vision for AI is inevitable," she said at the briefing, adding that the message is "get on board or get left behind." Her line on the other side: "We're building a national model for smart, sensible AI regulation, a model that puts communities above corporations and people above powerful technology."

How New York compares

New York is not alone. TechCrunch notes California's SB 53, in effect since January, requires frontier labs to have and publish safety frameworks, set up whistleblower protections and report critical safety incidents within 15 days. Illinois' SB 315 starts at the beginning of 2027 with similar requirements plus independent safety compliance audits.

Set side by side, the clocks differ sharply: 72 hours in New York, 15 days in California. For a lab operating in all three states, the shortest clock effectively becomes the company standard, because an incident that triggers one state's duty will often trigger the others. We lay out the legal differences in more detail in our comparison of the RAISE Act and SB 53, and we track incident rules across jurisdictions in our AI incident reporting requirements tracker.

New York has also stacked other AI measures. TechCrunch notes an AI companion law already being enforced and a separate chatbot safety law for minors, signed in October, that takes effect June 1, 2027.

What we would watch

The first test is simple: who is on the registration list on January 2, 2027? The $500 million threshold sweeps in the best-known labs, but the revenue line also catches companies that rarely appear in safety debates. A registry with gaps would tell us the definition is doing real work.

The second test is the first public safety frameworks and transparency reports. The labs already publish safety documents voluntarily. If the state-mandated versions look the same as the voluntary ones, the law will have standardized disclosure without changing behavior. If they differ, with firmer thresholds or named triggers, regulators will have made the labs commit in ways they did not before.

The third test is the first 72-hour report. A short clock invites disputes about when the clock started and what is "critical." Whether New York publishes incident counts, even anonymized, will show how much sunlight the law is meant to create.

Our position: this is the most useful kind of AI law for a fast-moving field, because it asks for facts rather than predictions. It is also only as good as DIGIT's willingness to say what it finds. Registration is the easy part, and a public registry of covered labs would itself be useful information for researchers and journalists. The proof will come in what the state does with the first report it receives and whether the public ever hears about it.