New York bill would add independent audits to state AI safety law

For the chief risk officer of a New York bank

State Sen. Andrew Gounardes and Assemblymember Jordan Wright introduced a bill in New York on Thursday (Oct. 29) that would build on the state's RAISE Act by adding independent safety audits, incident disclosure requirements and stronger enforcement, PYMNTS reported.1

According to PYMNTS, the RAISE Act only requires AI developers to disclose their safety plans, and the new bill adds oversight of whether developers follow them.1 Sen. Gounardes said in a press release that "we can't trust Big Tech companies to regulate themselves" when it comes to AI.1

The bill would require large developers to retain an independent, state-licensed "frontier examiner" to conduct annual audits of compliance with their safety frameworks and to make catastrophic risk assessments every 90 days, PYMNTS reported.1

Large developers would have to disclose any substantial jump in risk or in their models' ability to automate AI research within 72 hours, and report quarterly on how much of their AI research is automated and how they use computing power.1

The bill would let the newly created state Office of Digital Innovation, Governance, Integrity and Trust (DIGIT) limit the computing power large developers devote to training new models or to automated AI research.1

It would also empower the Attorney General to ask a court to halt dangerous activity, including by restricting a model's deployment in New York; the source did not say what penalties or timelines apply.1

Why it matters

If enacted, New York would supervise large AI developers in a way that resembles bank supervision, with outside examiners and a state office with real powers. A court order restricting a model's deployment in New York would act on the product itself, so banks that rely on these vendors should check whether their contracts cover a single-state restriction, not only outages or data center failures.

From the brief

Our take. Your AI vendor may soon get what your bank already has: an outside examiner. It may also face something your vendor plans probably skip: a court order restricting its model in New York, where you operate.

Action. Third-party risk: Add a regulator-ordered New York outage for each critical AI vendor to exit plans; request frontier-examiner audit results once available.

From our records

Sources

  1. ReportPYMNTS | New York Bill Would Raise AI Safety Regulations PYMNTS