New York Laws “RAISE” the Bar in Addressing AI Safety: The RAISE Act and AI Companion Models

Published
Score
1

Why it matters

On December 19, 2025, New York Governor Kathy Hochul signed the Responsible Artificial Intelligence Safety and Education (RAISE) Act into law, imposing safety, transparency, testing, and incident-reporting requirements on developers of "frontier" AI models—highly capable systems posing risks of critical harm, such as over 100 deaths or $1 billion in damage.[1][2][3][4][5] Key provisions mandate large developers to create and publicly disclose written safety and security protocols before deployment, including cybersecurity controls, ongoing testing, annual reviews, and designation of a senior compliance officer; prohibit deployment of models with "unreasonable risk of critical harm"; and require reporting safety incidents (e.g., autonomous harmful behavior) to the state within 72 hours.[1][3][4][6] The law establishes an oversight office in the New York Department of Financial Services for evaluation, fees, rules, and annual reports, with the Attorney General authorized to enforce via civil penalties up to $1 million for first violations and $3 million for subsequent ones, plus whistleblower protections.[1][3][5][6]

Involved parties include Governor Kathy Hochul, who signed the amended bill after negotiations; bill sponsors State Senator Andrew Gournardes and others (S6953B/A6453B); the state legislature, which passed it in June 2025; the Department of Financial Services (new AI oversight office); and the Attorney General for enforcement.[1][3][5][7] It targets "large developers" of frontier models, without naming specific companies, though implications apply to AI firms like those developing foundation or generative systems.[4][9]

The Act addresses AI risks like self-replication, deception, biological weapon design, and lack of federal regulation, building on New York's 2025 AI leadership and aligning with California's SB 53 amid federal delays.[1][3][5][8] Timeline: Passed by legislature June 2025; signed December 19, 2025 (with agreed amendments pending January 2026 publication); effective January 1, 2027 (or July 1 per some reports).[1][2][9]

Newsworthy for positioning New York as a leader in state-level AI safety regulation, harmonizing with California to pressure federal action, and imposing novel duties on frontier AI developers amid testing revelations of existential risks—all as models advance rapidly without unified U.S. oversight.[3][5][10] Recent analyses (January 2026) urge companies to prepare compliance now.[4][6]

Sources

mail

Get notified about new Law And Technology developments

Primary sources. No fluff. Straight to your inbox.

See more entries tagged Law And Technology.