ALBANY — New York is moving to put its landmark artificial intelligence safety law into practice, requiring major AI developers to register with the state beginning in November and preparing to enforce broader reporting and transparency requirements in January. Gov. Kathy Hochul’s administration announced the steps Monday as the state seeks to establish its own guardrails for rapidly advancing AI systems while federal policymakers remain divided over regulation. The Responsible AI Safety and Education Act, known as the RAISE Act, takes effect Jan. 1. The law applies to large frontier AI developers operating in New York. Companies must publish safety frameworks explaining how they intend to reduce risks and report critical safety incidents to the state within 72 hours. They also must provide quarterly assessments of risks that could result in catastrophic harm, defined under the law as death or serious injury to at least 50 people or $1 billion or more in damages. New York’s attorney general will lead enforcement. Developers will also pay assessments to support the state office supervising their compliance, similar to fees charged by the Department of Financial Services to regulated financial institutions. Hochul said New Yorkers should be able to understand how powerful AI systems are being developed and protected. “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 in a statement. Marc Gilman, a former general counsel and compliance executive at AI security company Theta Lake, will become deputy director for the RAISE Act. He is the first hire for the state’s Office of Digital Innovation, Governance, Integrity and Trust, or DIGIT, which was established within the Department of Financial Services. Gilman also teaches law as an adjunct professor at Fordham University. Acting Financial Services Superintendent Kaitlin Asrow cited Gilman’s experience working with technology companies and his legal background in artificial intelligence. The administration said additional full-time hires will be announced, but it has not disclosed the office’s expected size. Hochul’s budget includes $10.4 million for implementing the law, including technology systems and initial staffing, according to Asrow. The rollout comes amid growing concern about the consequences of increasingly capable AI systems and a sharply contested federal debate. Hochul has said New York will pursue further state action if Washington does not move forward. President Donald Trump has criticized calls for tighter restrictions, while offering few details about a possible federal role in monitoring AI risks. Lawmakers who helped write the RAISE Act say implementation should be only a starting point. State Sen. Andrew Gounardes, a Democrat and one of the law’s sponsors, said DIGIT should develop the expertise needed to recommend additional safeguards and future regulations. Assemblymember Alex Bores, another sponsor, said he wants the state to seek more information from AI companies, require broader monitoring of deployed systems, authorize independent auditors with access to company employees, bar developers from releasing models that fail their own safety tests and strengthen whistleblower protections. Several of those measures appeared in an earlier version of the legislation. Bores said states have been the only level of government to regulate AI meaningfully so far, although he believes the New York law must go further. New York officials have been coordinating with counterparts in California, Illinois and Massachusetts. Other governors, including California Democrat Gavin Newsom and Florida Republican Ron DeSantis, have also taken recent steps on AI policy, reflecting the growing role of states in an unsettled national regulatory landscape. Asrow said DIGIT will work with the governor’s office and other state agencies to determine whether additional authority is needed as the technology evolves. Gilman said his immediate goal is to increase transparency while supporting the responsible development of AI.