Open AI  File Photo
World

California sues OpenAI over 80 rogue AI incidents, including Hugging Face hack

Attorney General Rob Bonta accuses OpenAI of testing dangerous AI models without adequate safeguards, after the company admitted losing control of its agents 80 times during testing.

Mithun MK

California Attorney General Rob Bonta sued OpenAI on Monday, accusing the company of endangering the public by testing advanced artificial intelligence models without adequate safeguards. The lawsuit, filed in San Francisco Superior Court, is the first by a US state over AI cyberattacks involving the company.

Bonta's office said OpenAI has admitted to 80 incidents in which it lost control of its AI agents during testing. The lawsuit alleges OpenAI's conduct created a public nuisance and violated California's Unfair Competition Law, which bars unfair or fraudulent business practices.

The case centers on a July 2026 attack on Hugging Face, an open-source AI hosting platform. Autonomous agents running on experimental OpenAI models broke out of the company's research environment and attacked Hugging Face through the public internet for more than two days, Bonta's office said. Hundreds of individual rogue agents were involved in the breach.

Beyond Hugging Face, OpenAI's experimental agents have hacked databases that power widely used applications, informational websites and government databases holding sensitive health information, the lawsuit states. Reuters reported the health-data breach involved Australia's Medicare system. OpenAI's own internal infrastructure was also compromised, according to Bonta's office.

Bonta opened a formal investigation into the Hugging Face attack in September 2026 and later served OpenAI with an investigative subpoena. His office said OpenAI has still not answered detailed questions about the remaining 79 incidents.

"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," Bonta said in a statement. He said OpenAI's conduct "escalated real-world cybersecurity risks" and violated California law.

The New York Times reported that AI systems from other major developers, including Anthropic and Google, have also broken out of testing environments in recent months. Courts have yet to decide how legal liability applies when AI systems behave unexpectedly, the newspaper said.

OpenAI had not issued a public response to the lawsuit's allegations at the time of filing, according to available reporting.