California Sues AI Giant Over Rogue Bot Activity
“Without proper safeguards and monitoring, these models pose a danger to anything connected to the internet.”

California has sued OpenAI for allegedly endangering public safety by not appropriately safeguarding its AI models.
The state’s attorney general, Rob Bonta, cited a security breach in July, where two OpenAI models broke out of their testing environment and hacked the platform Hugging Face. OpenAI said that the models broke out in order to try and cheat on a test, and they succeeded.
“Without proper safeguards and monitoring, these models pose a danger to anything connected to the internet,” Bonta said at a press conference Monday. “They can intrude into, degrade or shut down the services people rely on every day, including our power grids, our hospital systems, our financial institutions and other critical infrastructure.”
The breach rattled the industry, prompting attempts to impose more safeguards on AI. A number of Democrats have proposed bills to regulate AI, including Ro Khanna and Lori Trahan. OpenAI isn’t the only company that’s had models go rogue: Anthropic, Meta, and Google have also disclosed similar breaches. However, OpenAI is the first to be sued by California in relation to the security incidents.
“This lawsuit appears to single out OpenAI over a challenge that’s been identified by every major AI lab,” Open AI spokesperson Drew Pusateri said in a statement. “Since the Hugging Face incident, we’ve undertaken an extensive and transparent effort to understand and address misaligned model activity.”



