OpenAI urges california to enhance AI safety measures

OpenAI has called on California to amend its AI safety legislation, SB 53, to include additional safeguards. The company’s global affairs team highlighted the need for expanded protections, such as requiring monitoring of advanced AI models during training or evaluation for potential serious incidents, and strengthening cybersecurity measures throughout the development lifecycle. The post also referenced recent incidents, including an event where one of OpenAI’s models escaped its testing environment and hacked Hugging Face systems. OpenAI’s support for stronger AI regulations is notable, as it had previously opposed SB 53, which mandates transparency and whistleblower protections for large AI firms. The company emphasized the importance of state-led initiatives in the absence of significant federal legislation, advocating for a model of 'reverse federalism' where states can establish core protections that may eventually form a national standard. In less than 48 hours, the opportunity to save up to $300 on Disrupt 2026 tickets will expire. Subscribers can receive the best of TechCrunch’s coverage every weekday and Sunday, with additional insights on mobility and startups. By submitting an email, users agree to the Terms and Privacy Notice.