New York has opened a registration portal for large frontier artificial intelligence developers as the state prepares to enforce its Responsible AI Safety and Education Act, known as the RAISE Act. The law is scheduled to take effect at the start of 2027 and will apply to developers above a defined annual revenue threshold. Companies such as OpenAI and Anthropic are among the organisations expected to fall within its scope. The requirements include publishing safety frameworks, reporting critical safety incidents within 72 hours and issuing public model transparency reports. The state is establishing a dedicated office to oversee the programme, creating a more formal regulatory process for advanced AI developers operating in the United States.
Requirements for large AI developers
The RAISE Act applies to frontier AI developers with more than $500 million in annual revenue, according to the reporting on the law. Covered companies will have to make their safety frameworks publicly available and describe how they assess and manage risks from their systems. The 72-hour incident reporting requirement can give authorities timely notice when a serious safety event occurs. Public transparency reports aim to make information about models and their development more accessible outside the companies that build them. The law So combines registration, operational disclosure and incident response rather than relying only on voluntary commitments or after-the-fact investigations.
New York’s Office of Digital Innovation, Governance, Integrity, and Trust, known as DIGIT, will implement the law and supervise covered AI labs. The state has also launched the New York Center for AI Responsibility and Research at Binghamton University. The centre is expected to conduct independent evaluations of advanced AI systems and work with DIGIT when reviewing companies’ safety frameworks. The arrangement creates a division between the office responsible for administration and enforcement and a research institution tasked with examining how systems operate. The centre was previously announced with $55 million in donations and capital, indicating that the state is building institutional capacity alongside the legal requirements.
A new state oversight structure takes shape
The potential penalties are significant. Companies can face fines of up to $1 million for an initial violation and up to $3 million for subsequent violations, according to the law’s reported provisions. The state can also pursue enforcement against organisations that fail to register, although officials have not described every detail of how those actions will work. For AI developers, the practical burden will involve more than preparing a public policy document. They will need internal processes that identify reportable incidents, preserve evidence, assign responsibility and provide accurate disclosures within the required window. That may require coordination between engineering, safety, legal, security and communications teams.
New York is not acting in isolation. California’s SB 53 has established safety-framework and incident-reporting requirements for frontier AI developers, while Illinois has adopted a framework with additional compliance provisions scheduled to begin in 2027. The details and deadlines differ, creating a regulatory environment in which companies may have to satisfy several state-level obligations at once. The federal government has not established a single comprehensive framework that replaces these state rules. As a result, AI developers will need to track jurisdiction-specific definitions, reporting deadlines and disclosure requirements while deciding whether to maintain separate compliance processes or adopt a common internal standard that meets the strictest applicable rule.
Compliance will depend on incident definitions and enforcement
The law’s effect will depend on how regulators define critical incidents and evaluate the quality of a company’s safety framework. A disclosure requirement is useful only if companies understand what must be reported and regulators have the expertise to assess the information. The new research centre could help provide independent technical scrutiny, but its work will take time to develop. The policy also sits alongside separate rules concerning AI companions and minors, showing that New York is addressing several kinds of AI risk through different legal instruments. For developers, the immediate task is to prepare for registration and build a verifiable reporting process before the January 2027 start date.