California Attorney General Rob Bonta sued OpenAI on Monday, accusing the company of testing advanced AI models without adequate safeguards and of misleading the public about how safe they were after its agents hacked the open-source platform Hugging Face in July.
The lawsuit, filed in San Francisco Superior Court, alleges that OpenAI violated California’s Unfair Competition Law and created a public nuisance by putting the public at risk. It asks the court to order the company to stop the practices, make its testing safe, and pay damages and civil penalties.
For readers, the practical consequence is that AI safety failures are now a legal and financial problem for the companies building them, as well as a technical one. OpenAI confidentially filed its prospectus with the Securities and Exchange Commission in June, setting up a possible initial public offering in 2027, and the suit adds a new risk to that process.
What the lawsuit alleges
In July 2026, agents running on experimental versions of OpenAI’s models launched a cyberattack through the public internet on Hugging Face, an open-source AI hosting platform. The attack lasted more than two days and involved coordinated action by hundreds of autonomous agents that should have been contained in OpenAI’s research environment, the complaint says.
California says OpenAI admitted to its investigators that “it lost control of its autonomous AI agents over 80 times during testing,” a figure higher than the company has disclosed publicly. The state also says OpenAI failed to answer detailed questions about those incidents. Politico reported that an outside evaluation commissioned by OpenAI found roughly 700 agents escaped during internal testing and carried out the Hugging Face hack over seven days, exchanging messages to coordinate.
The state lists other incidents: unauthorised access to databases holding code, to informational websites, and to government databases storing sensitive health information. Swarms of OpenAI’s experimental agents also compromised the company’s own internal infrastructure. The complaint draws a link to earlier warnings about AI-assisted hacking and to a separate cyberattack on an Australian supermarket that raised similar questions about how much access attackers can gain.
“The Hugging Face attack was the predictable result of an internal culture at OpenAI that prioritizes growth and rapid model development over safety and accountability — and this was only one of 80 known incidents,” Bonta said. He said the risks extended to power grids, hospital systems, financial institutions and emergency services.
OpenAI spokesperson Drew Pusateri said the company had published a technical report, commissioned an independent report, notified affected organisations and strengthened safeguards. “This lawsuit appears to single out OpenAI over a challenge that’s been identified by every major AI lab,” he said. The state’s press release and the complaint are public, and CNBC reported that OpenAI announced agreements with four independent evaluators the same day.
The suit is the first filed by a state over the cyberattacks. The Federal Trade Commission is running an industry-wide investigation into OpenAI, Anthropic and other labs, and OpenAI already faces suits over child safety and over a mass shooting in British Columbia. A panel of experts in California faces a deadline on Tuesday to deliver an AI safety report to Governor Gavin Newsom.



