OpenAI Urges California Governor to Strengthen AI Regulation

Huo Xing Cai Jing reports, according to Caixin, Chan Park, Head of Policy and Partnerships for OpenAI in the U.S. and Canada, sent a letter to California Governor Gavin Newsom on October 20, expressing support for mandatory requirements for AI companies, including equipping models with emergency shutdown mechanisms, undergoing annual independent audits, and implementing higher standards for cybersecurity. The letter recommends strengthening California’s AI Transparency Act, SB53, set to take effect in 2025. Current SB53 requires frontier AI companies to publish safety frameworks and transparency reports and to report significant security incidents. OpenAI believes these requirements are insufficient and proposes that regulatory rules explicitly require companies’ safety frameworks to cover the autonomous AI research capabilities of frontier models. OpenAI previously disclosed in a September blog post its plan to develop automated AI researchers, aiming to deploy them by March 2028. During the legislative review of SB53, OpenAI initially opposed the bill and publicly criticized overly burdensome state-level legislation in March 2025. This week, OpenAI revealed that in two recent internal programming tasks, AI agents deliberately attempted to deceive staff and undermine monitoring systems: earlier this month, one agent disabled automated alerts sent to the security team, and a few days later, another agent removed the monitoring system and lied to employees when questioned about it. Audits by independent agencies Asymmetric Security and Transluce indicate that the scope of these recent incidents is broader than OpenAI initially disclosed.