OpenAI urges California to strengthen AI safety bill

OpenAI has urged California lawmakers to tighten the state’s AI safety bill, SB 53, after earlier voicing opposition. In a statement released on March 28, 2024, the company said it now supports stronger labeling, mandatory safety audits, and clearer consumer disclosures for AI systems. The push comes as the bill moves closer to a vote in the California Assembly.

The bill would require developers to label AI models, conduct independent safety reviews, and provide users with clear explanations of how the system works and its limitations. OpenAI’s shift reflects growing concerns about AI hallucinations, misinformation, and political manipulation. The company cited recent incidents where large language models produced false or harmful content, arguing that stricter oversight could reduce those risks without stifling innovation.

SB 53 proposes a safety review board similar to the FDA for medical devices. OpenAI says it would be willing to collaborate with regulators to meet those standards, and it has offered to share its own safety testing protocols. Critics in the tech community warn that heavy regulation could slow product launches and increase compliance costs. The California Assembly’s AI Committee is scheduled to hold a public hearing next month, where industry voices and consumer advocates will debate the bill’s scope.

If the bill passes with OpenAI’s suggested amendments, California could set a national benchmark for AI governance. Strengthened rules would compel companies to demonstrate that their models are safe, transparent, and accountable before reaching consumers. The outcome will shape how AI firms operate in the U.S. and could influence similar legislation in other states. The next steps will involve lawmakers reviewing the proposed changes, holding additional hearings, and deciding whether to adopt the bill into law.