The California Assembly passed the AI Safety Bill SB 1047 on August 27, it now requires a final confirmation vote in the state Senate before Governor Newsom can sign it into law.
There is a global consensus on the need for AI regulation to address the growing risk of AI misuse, as shown by the recent launch of the EU AI Act and the African Continental AI Strategy.
As the global tech hub, California’s AI governance approach impacts the broader AI ecosystem prompting California State Senator Scott Wiener a Democrat to introduce the Safe and Secure Innovation for Frontier Artificial Intelligence Models Bill.
California’s Senate Bill 1047 (SB 1047) focuses on regulating frontier models, which are advanced AI systems that require a lot of computing power (10^{26} integer operations) and as a result, cost more than $100 million to train.
SB 1047 requires all frontier models to undergo safety tests, including a kill switch, be subject to third-party auditing, and the state attorney general can sue non-compliant developers.
The bill has sparked controversy, with former House Speaker Nancy Pelosi calling it “well-intentioned but ill-informed,” while Elon Musk supports it as necessary.
Developing AI regulations is a balancing act, aiming to foster innovation and maximize the benefits of AI while mitigating the risks associated with advanced AI models.
