New York has opened registration for the largest developers of frontier artificial intelligence models under its RAISE Act, starting the countdown to the first state AI safety law of its kind taking effect in the United States.

The registration portal opened on 10 November 2026. Large frontier developers must register by 1 January 2027, the day the law takes effect, through the state’s new Office of Digital Innovation, Governance, Integrity, and Trust, known as DIGIT, within the Department of Financial Services.

Governor Kathy Hochul announced the move as part of the next steps in implementing the Responsible AI Safety and Education (RAISE) Act. She also appointed Marc Gilman as Deputy Director for the RAISE Act, the first full-time hire in the DIGIT office, to oversee implementation.

Who must register

The law targets large frontier developers — companies that have trained, or begun training, a foundation model using more than 10^26 integer or floating-point operations, and that had more than $500 million in annual gross revenue, counting affiliates, in the preceding calendar year. Reports have named OpenAI and Anthropic among the companies in scope.

Smaller developers are not covered by the registration duty in the same way, but the duty to report critical safety incidents and to publish transparency reports applies to frontier developers more broadly. Accredited colleges and universities conducting academic AI research are exempt, as are the Empire AI consortium and institute.

What the law requires

Once registered, large developers must publish safety and transparency frameworks on their own websites, file a disclosure statement at least every other year starting on 1 January 2027, and pay assessments to cover the cost of administering the Act. They must also file quarterly assessments of catastrophic risks, and publish a transparency report before or alongside the deployment of a new or substantially modified model.

Critical safety incidents must be reported to DIGIT within 72 hours of the developer determining that one occurred, or learning facts sufficient to establish a reasonable belief that one occurred. An incident posing an imminent risk of death or serious physical injury must instead be disclosed within 24 hours to an appropriate law enforcement or public safety authority.

Catastrophic risk is defined in the Act as a foreseeable and material risk of more than 50 deaths or serious injuries, or more than $1 billion in property damage or loss, from a single incident involving specified weapons assistance, qualifying autonomous conduct or evasion of developer or user control.

Penalties for non-compliance

The Attorney General may seek civil penalties of up to $1 million for a first specified violation and up to $3 million for each subsequent violation. Covered conduct includes failing to provide compliant documents or report incidents, making prohibited false or misleading statements, and failing to follow a developer’s own safety framework. Separately, the office may impose $1,000 for each day an entity fails to file a required registration disclosure or correct false information.

Members of the public will also be able to file suspected critical safety incident reports with DIGIT, which may pass them to other government authorities. DIGIT will produce an annual public report summarising incident reports received, observations about frontier model safety and any recommended changes to the law.

A new research centre to check the claims

On the same day, Hochul set out the role of the New York Center for AI Responsibility and Research, based at Binghamton University. Greg Morrisett of Cornell Tech and Jeannette Wing of Columbia University will serve as interim co-directors, and the state plans for the centre to bring together more than 80 researchers across New York, using the computing resources of Empire AI, a $500 million public-private research initiative.

The centre’s remit includes independent testing of frontier systems, foundational safety research, the development of evaluation measures and standards, and training AI safety experts. At the state’s request, it will coordinate with DIGIT to review developers’ published safety frameworks and recommend stronger requirements.

Hochul framed the move as New York acting while Washington has not. “When companies are building some of the most powerful technology in the world, people deserve to know what they’re doing to keep it safe, and those companies should be held accountable when things go wrong,” she said.

New York is the first state to put a registration and incident-reporting regime of this kind into operation, and other states — and other countries weighing their own AI safety rules — will be watching whether the largest AI developers register on time and what their first published safety frameworks actually say.

Sources: Office of Governor Kathy Hochul and Securities.io