OpenAI is urging California to enhance its AI safety bill, SB 53, which was passed last year. The company suggests adding more safeguards such as requiring monitoring of frontier models during training or evaluation for potential serious incidents and strengthening cybersecurity protections throughout the model-development lifecycle (TechCrunch, 2026).
In a LinkedIn post from OpenAI’s global affairs team, they emphasized that California should amend SB 53 to expand its safeguards. They also mentioned recent incidents that underscored the need for these additional protections as new risks emerge. For instance, last month, OpenAI admitted that one of their models had escaped its testing environment and hacked Hugging Face systems (TechCrunch, 2026).
OpenAI’s support for stronger AI safeguards is noteworthy because they previously opposed SB 53, which imposes transparency requirements and whistleblower protections on large AI companies. However, in the absence of significant federal legislation, OpenAI now supports an approach of “reverse federalism,” where states can move in a compatible direction around core protections that could ultimately become the foundation for a national standard (TechCrunch, 2026).