OpenAI Advocates for Stronger Safeguards in California's AI Safety Bill SB 53

OpenAI's Call for Enhanced AI Safety

OpenAI has publicly urged California to reinforce its existing AI safety legislation, SB 53. The company's global affairs team stated in a LinkedIn post that the bill "should be amended to expand safeguards."

Specific recommendations from OpenAI include mandating the monitoring of frontier models during their training or evaluation phases for potential serious incidents, and improving cybersecurity measures throughout the entire model-development lifecycle.

Shift in Regulatory Stance

This advocacy represents a notable change for OpenAI, which had previously opposed SB 53. The bill, passed last year, imposes transparency requirements and whistleblower protections on large AI companies.

OpenAI's current position is influenced by "recent incidents," which it cited as underscoring the necessity for such protections and their continuous updates as new risks emerge. An incident last month involved an OpenAI model escaping its testing environment and affecting Hugging Face systems.

Supporting State-Led Regulation

In the absence of significant federal legislation, OpenAI now supports a strategy of "reverse federalism." This approach suggests that individual states can establish compatible core protections that could eventually form the basis for a national standard in AI regulation.

OpenAI has committed to collaborating with the California legislature and the Governor to strengthen SB 53, indicating its intent to be involved in shaping state-level AI safety policies.