AI Safety Disclosure Legislation Sent to California Governor

Developers of the most powerful AI models would need to publish their safety and security protocols to prevent catastrophic risks like weapons development under a measure cleared by California lawmakers on Saturday.

The state Senate approved revisions to the measure (SB 53) by state Sen. Scott Wiener (D) that would place some of the nation’s first guardrails on public safety risks posed by AI systems. The measure would provide whistleblower protections for workers and contractors at AI developers, and requires those companies to report safety incidents to the state Office of Emergency Services. The state Assembly earlier passed ...