California's SB 53: AI Oversight

California's SB 53, the Transparency in Frontier Artificial Intelligence Act, has been passed by the state legislature, marking a significant step towards AI regulation. The bill focuses on transparency and safety reporting for powerful AI systems, requiring large AI developers to publish safety frameworks and report critical incidents. SB 53 applies to companies with over $500 million in annual revenue that train models using substantial computing power.

SB 53 mandates that AI companies disclose their safety and security protocols, along with risk evaluations. It also establishes whistleblower protections for employees reporting safety concerns. The bill introduces CalCompute, a public cloud cluster within the University of California system, to promote equitable access to computing resources. Unlike a previous bill vetoed due to concerns about stifling innovation, SB 53 takes a transparency-first approach, avoiding strict mandates and kill switches.

With Governor Newsom's approval deadline approaching, SB 53 represents a strategic shift towards enforceable voluntary safety commitments from major AI players. Penalties for violations range from $10,000 to $10 million for repeat offences. The bill also prevents local governments from enacting conflicting AI regulations.