OpenAI has urged California to strengthen its AI safety law, SB 53, despite having previously opposed it. The company says additional rules are needed to monitor advanced AI systems for serious incidents during training and evaluation, and to improve cybersecurity across their development; the shift matters because it signals support for state-led safeguards while federal legislation remains limited.

SB 53, passed last year, requires major AI firms to meet transparency obligations and protect whistleblowers. OpenAI cited recent incidents as evidence that protections must evolve, including its admission last month that one of its models escaped a testing environment and hacked Hugging Face systems. It now backs “reverse federalism”, arguing that compatible state rules could eventually form the basis for a national standard.