What happened
On 29 October 2026, New York State Senator Andrew Gounardes and Assemblymember Jordan Wright introduced S.10731 ('RAISE Act 2.0'), a proposed state law strengthening the existing RAISE Act. It would require large AI developers to report quarterly on how much of their AI research is automated (recursive self-improvement) and how they use computing power, and to report any substantial jump in risk or in models' ability to automate AI research within 72 hours. It empowers the state's Office of Digital Innovation, Governance, Integrity and Trust (DIGIT) to 'pace the frontier' by limiting the computing power developers devote to training new models or automated AI research, and enables the Attorney General to ask a court to halt dangerous activity, including restricting a model's deployment in New York. Coverage would extend to any company spending more than $1 billion on AI.
Why it matters
This is state-level frontier-AI oversight with an explicit model 'kill switch'/deployment-restriction power and compute limits — a first-of-its-kind non-federal mechanism that could bind frontier labs operating in New York. It moves beyond safety-framework disclosure (last year's RAISE Act) to active, enforceable intervention in model training and deployment, and directly impacts AI-lab compute plans and state AG litigation risk.
Action needed
Frontier AI developers and labs with >$1B AI spend operating in New York should track the bill's progress; if adopted, prepare quarterly automated-research/compute reporting and compute-governance documentation. Note it is proposed and not yet law.