What CIOs should know about California’s contentious AI bill

California is gearing up to establish AI guardrails across the state through its Safe and Secure Innovation for Frontier Artificial Intelligence Models Act, which was recently passed by the state Assembly and is awaiting Governor Gavin Newsom’s signature.

The bill primarily focuses on regulating the development of AI models, mandating providers to adhere to various protocols, including:

• Implementing a capability for complete shutdown of the model
• Retaining an unredacted copy of safety and security protocols
• Submitting a statement of compliance to the state’s attorney general
• Providing whistleblower protections

Entities investing more than $10 million in refining an existing model will be held accountable as developers under the regulations. Failure to comply may lead to legal action from the state’s attorney general.

The act also paves the way for California to establish a separate board dedicated to AI oversight, updating definitions, issuing regulations, and fostering responsible AI deployment.

While technology leaders are monitoring the evolving landscape of AI regulations, many are acknowledging their organizations’ shortcomings in scaling responsible AI practices, citing concerns over potential regulatory delays.

Despite the overall support for AI regulation from technology providers, some have expressed reservations about perceived overreach by California lawmakers, with concerns about the impact on AI innovation in the state.

Senator Scott Wiener, the bill’s sponsor, emphasized the necessity of SB 1047, highlighting the collaborative efforts to enhance the bill’s provisions.

Leading up to SB 1047’s implementation, CIO Dive outlines key enforcement details that technology leaders should be aware of:

When will the act become law?

Following a final confirmation vote, Governor Newsom will have until Sept. 30th to sign the bill into law or veto it.