New York’s RAISE Act Brings New Oversight to Frontier AI Companies

Preface

New York is moving to enforce one of the country’s most significant AI safety laws. The state has opened a registration portal for large frontier AI developers, including Anthropic and OpenAI, which must register by January 1, 2027. The RAISE Act introduces requirements for safety frameworks, public transparency, and timely reporting of critical incidents. Its implementation will be overseen by the newly formed Office of Digital Innovation, Governance, Integrity, and Trust (DIGIT), alongside an independent research center based at Binghamton University. Together, these measures create a broader system for monitoring advanced AI development, while adding to New York’s existing rules for companion chatbots and protections for minors. Other states, including California and Illinois, have also adopted or are preparing similar requirements.

Lazy bag

New York’s RAISE Act applies to large frontier AI developers with more than $500 million in annual revenue. It requires covered companies to register, publish safety frameworks and public model transparency reports, and report critical safety incidents within 72 hours. DIGIT will administer the law, with penalties reaching $1 million for one violation and $3 million for subsequent violations. A new center at Binghamton University will independently evaluate advanced AI systems and work with DIGIT. The state’s broader rules also address companion chatbots and minors, while California and Illinois pursue related safety and transparency measures. The central shift is from voluntary assurances toward formal disclosure, oversight, and enforcement.

Main Body

New York is preparing to put a comprehensive AI oversight law into practice. The state opened a registration portal on Tuesday for large frontier AI developers, establishing a formal process for companies covered by the RAISE Act. Developers such as Anthropic and OpenAI will be required to register by January 1, 2027. The law takes effect at the start of the year and sets out obligations intended to make the safety practices of major AI developers more visible to regulators and the public.

The RAISE Act applies to large frontier developers—companies with more than $500 million in annual revenue. Covered companies must publish safety frameworks on their websites, report critical safety incidents within 72 hours, and issue public model transparency reports. These requirements combine advance disclosure about how a company approaches safety with a process for notifying authorities when serious incidents occur. The public reporting provisions are intended to give people outside the companies, including regulators and researchers, more information about the development and operation of powerful AI systems.

The law’s requirements place emphasis on both preparation and response. Publishing a safety framework asks a developer to describe its approach before a reported incident becomes the focus of public attention. The 72-hour incident reporting deadline, meanwhile, establishes a specific period for communicating critical safety events. Public model transparency reports add a further disclosure channel. Taken together, these duties are designed to provide a clearer account of how covered labs assess risks, respond to problems, and explain their models and policies.

Administration will fall to the newly formed Office of Digital Innovation, Governance, Integrity, and Trust, known as DIGIT. The office will implement the RAISE Act and oversee large frontier AI labs. New York can impose penalties of up to $1 million for one violation and $3 million for subsequent violations. The state can also bring enforcement actions against companies that fail to register through the new portal. A spokesperson declined to provide specifics about those actions, leaving the precise enforcement process less fully described in the announcement.

Governor Kathy Hochul framed the law as a response to arguments that rapid AI development should take precedence over other concerns. Speaking at a press briefing on Tuesday, she said, “Tech moguls tell us that their vision for AI is inevitable. They tell us get on board or get left behind… That any collateral damage is worth it as long as we win the race to the AI cliff.” Her remarks presented the legislation as an effort to ensure that public safety and community interests remain part of decisions about AI development.

On Tuesday, Hochul also launched the New York Center for AI Responsibility and Research at State University of New York’s Binghamton University. The center will independently evaluate the safety, security, and operation of advanced AI systems. It will also work with DIGIT to review companies’ safety frameworks, creating a research role alongside the state office’s administrative and oversight functions. Hochul first announced the center in January with $55 million in donations and capital.

The center’s role is distinct from the reporting obligations imposed on companies. Developers must produce and publish specified information under the law, while the center is intended to assess advanced systems independently and contribute to review of safety frameworks. This arrangement could give state oversight a research component as well as a regulatory one. The announcement does not provide further details about the center’s evaluation methods, the frequency of its reviews, or how its findings will be published.

Hochul described the broader approach as an effort to establish a model for other jurisdictions. “We’re building a national model for smart, sensible AI regulation, a model that puts communities above corporations and people above powerful technology,” she said. The state’s stated position is therefore not limited to registration and company disclosures: it presents oversight as a way to balance technological development with public interests. How the framework operates in practice will depend on implementation, the scope of regulatory actions, and how companies meet their obligations.

The RAISE Act is part of a wider set of AI safety rules already adopted in New York. Last year, the state began enforcing a law for AI companion operators. It requires them to regularly tell users that they are speaking with a bot and to make reasonable efforts to detect suicidal ideation and self-harm. These requirements focus on user-facing interactions and potential harm to individuals, rather than the broad organizational safety and transparency disclosures required of frontier developers under the RAISE Act.

New York has also enacted a separate chatbot safety law for minors. Hochul signed it in October, and it is set to take effect June 1, 2027. The law restricts behaviors including deceptive human-like interactions, emotionally manipulative or sycophantic responses, and encouragement of behaviors such as suicide or disordered eating. Its focus on young users and chatbot behavior complements the companion-chatbot law, while the RAISE Act addresses the responsibilities of large developers and their advanced systems.

Other states are pursuing related approaches to AI safety and transparency. California’s SB 53 took effect in January. It requires frontier AI labs to have and publish safety frameworks, establish whistleblower protections, and report critical safety incidents within 15 days. The incident reporting period differs from New York’s 72-hour requirement, while both laws require safety frameworks and incident reporting. California’s inclusion of whistleblower protections adds another mechanism for raising concerns within AI organizations.

Illinois’ SB 315 is scheduled to take effect at the start of 2027. It contains similar requirements and also calls for independent safety compliance audits. The audit provision gives Illinois’ framework a specific independent review element, while New York’s announcement describes the Binghamton University center as an independent evaluator that will work with DIGIT. These state laws share a concern with making frontier AI safety practices more accountable, but they differ in their deadlines and particular oversight tools.

As these laws take effect, AI developers operating across state lines may face overlapping obligations. New York requires covered companies to register, publish safety and transparency information, and report critical incidents within 72 hours. California specifies a 15-day incident reporting period and whistleblower protections, while Illinois adds independent compliance audits. The practical implications will depend on each state’s implementation and enforcement, as well as which companies meet the applicable thresholds and definitions.

New York’s approach combines company-level duties, state enforcement, and independent research capacity. The RAISE Act establishes the main requirements for large frontier developers, DIGIT is responsible for oversight, and the Binghamton University center is tasked with evaluating advanced systems and reviewing safety frameworks with the office. Alongside the state’s separate rules for companion chatbots and minors, these measures form a layered regulatory approach. Its effectiveness will ultimately be judged by how consistently the requirements are applied and whether disclosure and review translate into meaningful safety improvements.

Key Insights Table