California’s Landmark AI Safety Bill SB 53 Advances—Newsom’s Decision Awaits
California Passes Major AI Safety Bill—Governor's Signature Still Pending
In a significant move for AI regulation, California's state senate has approved SB 53, a bill designed to set new transparency and safety standards for large artificial intelligence companies. The legislation now awaits a crucial decision from Governor Gavin Newsom, who has yet to comment publicly on the bill.
What Does SB 53 Require?
Authored by Senator Scott Wiener, SB 53 introduces several key requirements for AI companies:
- Transparency: Large AI labs must publicly disclose their safety protocols.
- Whistleblower Protections: Employees at AI labs will be granted new protections if they report unsafe practices.
- Public Cloud Access: The bill proposes the creation of CalCompute, a public cloud to broaden access to compute resources across the state.
How Is SB 53 Different from Previous Attempts?
Last year, Governor Newsom vetoed a broader AI safety bill from the same author, citing concerns about applying strict standards to all large models, regardless of their specific use cases. However, after feedback from a panel of AI experts convened by Newsom, SB 53 was amended to be more targeted:
- AI companies developing “frontier” models with less than $500 million in annual revenue will only need to provide high-level safety disclosures.
- Larger companies must submit more detailed safety reports.
Industry Response: Divided Opinions
SB 53 has sparked debate across the tech sector:
- Several Silicon Valley companies, venture capital firms, and lobbying groups oppose the bill, arguing it could lead to inconsistent standards if not aligned with federal or European regulations.
- OpenAI, in a letter to Governor Newsom, urged that companies meeting federal or EU safety rules should be considered compliant in California to avoid “duplication and inconsistencies.”
- Andreessen Horowitz’s legal team warned that bills like SB 53 may cross constitutional lines by interfering with interstate commerce.
- Anthropic, a leading AI safety company, publicly endorsed SB 53. Co-founder Jack Clark emphasized the need for federal standards but called SB 53 a "solid blueprint for AI governance that cannot be ignored."
What Happens Next?
The bill is now on Governor Newsom's desk. While he has previously recognized the importance of addressing AI risks, his final decision will be closely watched by both the tech industry and policymakers worldwide. If signed, SB 53 could serve as a model for AI regulation across the United States.
References
- California lawmakers pass landmark bill that will test Gavin Newsom on AI (Politico)
- Senator Scott Wiener on SB 53 (X/Twitter)
- SB 53 Bill Text (California Legislature)
- OpenAI Letter to Governor Newsom
- The Commerce Clause in the Age of AI (a16z)
- Anthropic Endorses California's AI Safety Bill SB 53 (TechCrunch)