The New York RAISE Act — the Responsible AI Safety and Education Act — is New York’s state law requiring large developers of advanced AI models to register with the state, publish safety protocols, and report serious safety incidents. From November 2026, frontier AI developers with over $500 million in annual revenue must register with the Department of Financial Services; full compliance and enforcement begin in January 2027.
What Hochul Announced on Monday
On Monday, September 21, 2026, New York Governor Kathy Hochul held a press conference in New York City to announce the next phase of the New York RAISE Act, the state’s RAISE Act AI safety law she signed late last year. Flanked by Attorney General Letitia James, Assemblymember Alex Bores, and Senator Andrew Gounardes — the lawmakers who carried the legislation — Hochul laid out how the law moves from paper to practice.
The headline: starting in November 2026, large frontier AI developers — companies at the scale of OpenAI or Anthropic, with more than $500 million in annual revenue — must register with the New York State Department of Financial Services (DFS) and begin preparing for full compliance. In regulatory terms, registration is the on-ramp. It tells the state who these companies are, what they build, and who is accountable for it.
The RAISE Act has been described as a nation-leading, first-of-its-kind state AI safety law, and Monday’s announcement was the first detailed look at its enforcement machinery since the signing. Hochul framed the moment as the point where the law stops being a promise and starts being a process.
What the Law Requires: November Registration, January Enforcement
First, some definitions. A “frontier AI developer” is a company building the most capable AI systems — the large models trained with enormous computing budgets that set the state of the art. Think of the labs behind the systems the public interacts with every day. The $500 million annual revenue threshold is designed to capture those players, not startups or small firms.
DIGIT — the Office of Digital Innovation, Governance, Integrity and Trust — is the new office inside the Department of Financial Services that will run the program. It is the state’s designated AI watchdog, handling registration, incident reporting, and ongoing oversight.
Here is the timeline and what each stage demands:
| Date | Milestone | What it means |
|---|---|---|
| November 2026 | Registration begins | Frontier AI developers (over $500M annual revenue) must register with the NY Department of Financial Services and begin preparing for compliance |
| January 2027 | Full compliance and enforcement | All requirements take effect; the state can begin enforcing against non-compliant companies, with civil penalties on the table |
And here is the substance — what covered companies must actually do, and what the law does not yet demand:
| The law requires | The law does not (yet) require |
|---|---|
| Publish safety protocols — the procedures a company follows to keep its models from causing harm | A mandatory AI kill switch — the governor said this is under active consideration, not yet mandated |
| Report critical safety incidents to the state within 72 hours | A ban on training or deploying advanced models |
| Regular reporting to the DIGIT office inside the Department of Financial Services | Federal-style pre-approval of new models before release |
| Registration with DFS starting November 2026, with civil penalties for non-compliance | Obligations for smaller developers under the $500M revenue threshold |
The 72-hour incident reporting rule is one of the law’s sharpest teeth. When a serious safety failure occurs — the kind of event documented in recent AI safety coverage, from OpenAI’s misalignment cases (covered here in our AI safety reporting) to Google’s Gemini testing breaches (see the Gemini breakout story) — covered companies must tell the state quickly. For readers accustomed to learning about AI incidents months later through leaks and investigations, that is a genuine change in how the industry discloses risk.
The Kill Switch Question
A “kill switch” in AI policy means an emergency mechanism to shut down or halt an AI system if it behaves dangerously — the software equivalent of the big red button on a factory floor. Until now, it has been mostly a thought experiment in AI safety circles.
On Monday, Hochul put it on the table for New York. She would not rule it out:
“We’ll look closely at additional ways to build on the RAISE Act and strengthen our AI laws… We may even explore safeguards like AI kill switches if they’re deemed feasible and in the best interest of our state.”
That line landed just three days after California Governor Gavin Newsom signed an executive order on September 18 exploring a mandatory AI kill-switch framework for state-regulated systems. Two of the country’s largest states are now, in the same week, openly discussing emergency shut-off requirements for advanced AI — a convergence that would have sounded like science fiction a year ago.
Hochul did not say how a New York kill-switch requirement would work, which systems it would cover, or when it might appear. The word “feasible” is doing heavy lifting in her statement: the technical question of whether a true kill switch can exist for a widely deployed model is unsettled, and the industry will push back hard on the practicality. For now, it is a live policy question, not a rule.
A Warning to the Industry: “Bring It On”
Hochul made it clear she expects a legal fight — and is ready for one. She said she knew some AI developers were already on the phone with their legal teams planning lawsuits against the state:
“Here’s my message to them: Bring it on. We will fight you in court, and we will hold you accountable.”
That is unusually combative language from a governor toward an industry that has grown accustomed to being courted. It reflects a real calculation: the RAISE Act’s $500 million revenue threshold deliberately targets companies with the deepest legal war chests in the sector, and everyone involved knows litigation over the law’s reach is coming.
The law gives the state civil penalties for non-compliance once enforcement begins in January 2027 — fines, not criminal charges. But the broader signal is the point: New York intends to treat AI oversight the way it treats financial oversight, through the same Department of Financial Services that polices Wall Street. The message to the industry is that the era of self-reported, self-policed safety is ending in the country’s fourth-largest state economy.
New York vs. Washington
Hochul framed the announcement as a direct rebuke of the federal government:
“Donald Trump and Washington Republicans may be standing still as AI grows more unpredictable, but New York will not.”
The contrast is real and deliberate. Trump has repeatedly dismissed AI safety fears as a “hoax,” and on Saturday, September 19 — two days before Hochul’s press conference — he announced an “AI Force” and a new AI czar aimed at accelerating the industry, not restraining it (full story on Trump’s announcement here). Where Washington’s posture is growth-at-all-costs, Albany’s is register, report, and prove your systems are safe.
This sets up the dominant regulatory storyline of the coming year: a federal government racing to grow the AI industry while the largest states build their own guardrails around it. California is exploring kill switches; New York is registering developers and demanding incident reports. The companies caught in the middle will have to navigate both.
The political backdrop matters too. Public anxiety about AI safety has been climbing: a September 16 transcript of Hochul’s remarks noted that alarm spiked after former Anthropic researcher Jacob Coxon warned that AI “could kill us all by the end of the decade” when he resigned. Hochul also pointed to New York’s one-year moratorium on hyperscale data centers — a separate, already-announced move — as evidence the state is willing to pump the brakes on AI infrastructure, not just software.
What Happens Next
The law now has an enforcement team taking shape. Hochul announced the appointment of Marc Gilman as Deputy Director for the RAISE Act — the first full-time hire of the new DIGIT office, which she had previewed in her State of the State address earlier this year. More full-time staff will join in the coming weeks and months, building the office that will receive incident reports and manage the registration process.
What to watch from here:
- November 2026 registration. Which companies register, and whether any challenge the $500 million threshold or argue they fall outside the definition of a frontier AI developer. The registration list itself will be a story.
- The lawsuits. Hochul has already invited the fight. Expect industry challenges to land before the January 2027 enforcement date, testing whether a state can impose these requirements on a nationally operating industry.
- The kill-switch exploration. California’s executive order and New York’s stated interest create a two-state front on emergency shut-off safeguards. Technical feasibility studies and industry reaction will determine whether this becomes real regulation.
- January 2027 enforcement. The moment the law’s penalties become live. Whether any company is penalized in the first enforcement cycle will say everything about how seriously the requirements are taken.
- The DIGIT office build-out. A regulator is only as strong as its staff. Gilman’s appointment is the first hire; the quality and independence of the team that follows will define how the law works in practice.
This is a developing story. The registration requirement is confirmed and dated; the enforcement mechanics and the kill-switch question are still being written. We will update this coverage as the November deadline approaches.
FAQ
What is the New York RAISE Act?
The Responsible AI Safety and Education (RAISE) Act is New York’s state law requiring large frontier AI developers — companies with over $500 million in annual revenue — to register with the state Department of Financial Services, publish safety protocols, and report critical safety incidents within 72 hours. It has been described as a nation-leading, first-of-its-kind state AI safety law. Registration begins November 2026; full enforcement starts January 2027.
Do AI companies have to register in New York?
Yes — if they qualify as large frontier AI developers with more than $500 million in annual revenue. Starting November 2026, those companies must register with the New York State Department of Financial Services and prepare for compliance, with full enforcement and civil penalties beginning in January 2027. Smaller developers below the threshold are not covered.
What happens if an AI company doesn’t comply with the RAISE Act?
Non-compliance carries civil penalties once enforcement begins in January 2027. Governor Hochul has also warned the industry she expects legal challenges and says the state will fight them in court: “Bring it on. We will fight you in court, and we will hold you accountable.”
Is there an AI kill switch in New York?
Not yet. Governor Hochul said on September 21, 2026, that New York “may even explore safeguards like AI kill switches” if they are deemed feasible and in the state’s best interest. It is under active consideration, following California’s September 18 executive order exploring a mandatory AI kill-switch framework — but no rule has been proposed.
Who is enforcing the RAISE Act?
Enforcement runs through the new Office of Digital Innovation, Governance, Integrity and Trust (DIGIT) inside the New York State Department of Financial Services. Governor Hochul appointed Marc Gilman as Deputy Director for the RAISE Act — the office’s first full-time hire — with more staff joining in the coming weeks and months.
References
- New York State Department of Financial Services, “AI Safety: Governor Hochul Announces Next Steps to Regulate Major AI Developers and Protect New Yorkers”, September 21, 2026
- New York Post, “AI companies will have to register in New York as Hochul won’t rule out a ‘kill switch’ for the tech”, September 21, 2026
- Washington Examiner, “New York orders frontier AI developers to register with state”, September 21, 2026
- USA Today Network (lohud), “New AI protections coming to NY as Hochul eyes ‘kill switch’”, September 21, 2026
- Law.com / New York Law Journal, “New York Begins Implementing RAISE Act as Hochul Signals More AI Regulation Ahead”, September 21, 2026
- EIN Presswire, “Video & Audio/Rush Transcript: Governor Hochul Discusses Keeping AI Developers Accountable”, September 2026
Leave a Reply