California Senator Revives AI Safety Reporting Bill After Previous Veto
SB 53 would require major AI companies to publish safety protocols
California State Senator Scott Wiener has introduced amendments to Senate Bill 53 that would require large artificial intelligence companies to publish safety protocols and report safety incidents. The legislation aims to create transparency requirements for major AI developers like OpenAI, Google, Anthropic, and xAI, making California the first state to implement such measures if passed. This follows Governor Gavin Newsom's veto of Wiener's previous bill, SB 1047, which contained similar provisions.
After vetoing the earlier legislation, Governor Newsom established a policy group led by AI experts to develop recommendations for the state's approach to AI safety. The group's final report emphasized the need for industry transparency requirements, which directly influenced the amendments to SB 53. The revised bill attempts to balance safety oversight with supporting California's growing AI industry, addressing concerns that led to the previous veto.
KEY POINTS
- •CA Senator revives AI safety bill
- •Bill requires safety reports from AI firms
- •Previous version vetoed by Gov. Newsom
The legislation includes whistleblower protections for employees who identify critical risks posed by their company's AI technology. It also proposes creating CalCompute, a public cloud computing resource to support AI startups and researchers. Unlike the previous bill, SB 53 does not make AI developers liable for harms caused by their models and is designed to avoid burdening smaller companies that utilize or fine-tune existing AI models.
Similar legislative efforts are underway in New York, where Governor Kathy Hochul is considering the RAISE Act, which would also mandate safety reporting from large AI developers. The viability of state-level AI regulation was recently affirmed when the U.S. Senate rejected a proposed 10-year moratorium on state AI laws by a vote of 99-1 in July 2025. This decision allows states to continue developing their own regulatory frameworks in the absence of comprehensive federal legislation.
The AI industry has shown mixed responses to transparency requirements. While Anthropic has expressed support for increased transparency, companies like OpenAI, Google, and Meta have demonstrated greater resistance. Currently, major AI developers publish safety reports voluntarily but with inconsistent timing and thoroughness. SB 53 now advances to the California State Assembly Committee on Privacy and Consumer Protection, with several legislative hurdles remaining before it could reach Governor Newsom's desk for consideration.