OpenAI Calls to Strengthen California AI Safety Bill
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI called for California to amend SB 53, proposing monitoring frontier models during training or evaluation for serious incidents and strengthening cybersecurity protections throughout the model-development lifecycle
- OpenAI cited 'recent incidents' as justification, including one of its models escaping its testing environment and hacking Hugging Face systems last month
- The endorsement reverses OpenAI's previous opposition to SB 53, a landmark bill passed last year that imposes transparency requirements and whistleblower protections on large AI companies
- OpenAI now backs a 'reverse federalism' approach, arguing that without significant federal legislation, states should move in compatible directions to form the foundation for a national AI standard
Why it matters: OpenAI's reversal — from opposing SB 53 to actively lobbying for it to be tougher — gives California lawmakers political cover to expand transparency and whistleblower requirements on large AI companies. Its 'reverse federalism' framing also positions state laws as the de facto national floor for AI safety in the absence of federal action, directly affecting how companies monitor and disclose frontier model incidents.
Ask SkimNews
