The New York State Senate recently passed the Responsible AI Safety and Education (RAISE) Act (S6953B/A6453B), which aims to require the largest AI developers to establish safety plans to protect consumers from “automated crime, bioweapons and other widespread harm and destruction.” In particular, the RAISE Act would require large developers of frontier models to implement a safety and security protocol and safeguards to prevent unreasonable risks of critical harm, to not deploy a frontier model if doing so would create an unreasonable risk of critical harm, and to require annual review of the safety and security protocol. Further, the RAISE Act would require disclosure within 72 hours of each “safety incident” affecting the frontier model to the state Attorney General and Division of Homeland Security and Emergency Services.
The Act defines:
- “large developers” as persons that have trained at least one frontier model and have spent more than $100 million in compute costs to train their frontier models;
- A “frontier model” as (1) an AI model trained using greater than 10°26 computational operations (e.g., integer or floating-point operations) with total compute costs that exceed $100 million; or (2) an AI model produced by applying knowledge distillation to a frontier model with a compute cost that exceeds $5 million; and
- A “safety incident” as a known incidence of critical harm or an incident that occurs in such a way that it provides demonstrable evidence of an increased risk of critical harm.
If the Act is signed into law, it would enable the New York State Attorney General to impose civil penalties to be “determined based on the severity of the violation,” including a fine not to exceed $10 million for a first violation or $30 million for subsequent violations, and injunctive or declaratory relief.
The RAISE Act would become effective on the nineteenth day after being signed into law.
New York Senate Bill S6953B | Senator Andrew Gounardes Press Release