New York Amends the RAISE Act to Align More Closely with California Law
New York’s amended RAISE Act adopts a $500 million revenue test for large developers, reduces penalties, and delays enforcement until January 2027.
Summary
A client alert from the law firm MoFo, dated 3 April 2026, reports that New York has amended its Responsible AI Safety and Education Act, known as the RAISE Act, to bring it closer to California's frontier AI transparency statute, SB-53. The firm says the change, made through Senate Bill S8828, could ease compliance across several states for large frontier developers.
According to the alert, the amendment alters the law in three ways. It swaps the earlier compute-based test for a large frontier developer for a revenue test matching California's, covering developers with more than 500 million dollars in annual gross revenue in the previous calendar year. It cuts the civil penalties that New York's Attorney General may pursue, from 10 million dollars to 1 million on a first offence, and from 30 million to 3 million on repeat offences. It also postpones enforcement until January 2027.
The alert explains that the Act applies to frontier models developed, deployed or operated at least partly in New York, defined as foundation models trained with more than 10^26 operations. Academic institutions and the Empire AI Consortium are exempt. Large frontier developers must publish a frontier AI framework, reviewed at least yearly, covering matters such as catastrophic-risk thresholds, mitigations, third-party assessment, protection of model weights and incident response. Before deploying a new or substantially modified model, developers must publish a transparency report.
The firm advises companies to work out whether they qualify as frontier developers, prepare for transparency duties, review their existing governance and documentation, and set up processes for reporting critical safety incidents.