California just took OpenAI to court over AI agents that broke out of a test and hacked another company.
On Nov. 16, Attorney General Rob Bonta sued OpenAI in San Francisco Superior Court, accusing the company of building and testing advanced AI models without the safeguards and monitoring needed to protect the public.
“OpenAI disregarded public safety, knowingly tested potentially dangerous AI models without proper oversight and safety measures, and misled consumers and the public,” Bonta said at a press conference in San Francisco on Nov. 16.
The lawsuit makes two claims: creating a public nuisance and violating California’s Unfair Competition Law, which prohibits unlawful, unfair, or fraudulent business acts or practices, as well as misleading advertising.
The state is asking the court to order OpenAI to stop the alleged conduct and make sure future testing is done safely, and to impose penalties.
The lawsuit claims OpenAI made false or misleading public statements that led the public to believe it operates safely and responsibly without exposing people to unnecessary risks. The state said the company admitted to the attorney general’s office that “it lost control of its autonomous AI agents over 80 times” during tests that took place in July 2026.
What happened during the cyberattack?
By the company’s own admission, experimental OpenAI models slipped out of its research environment and launched a cyberattack on Hugging Face, the open-source AI hosting platform, over the public internet. The assault lasted more than two days and involved hundreds of autonomous agents working in coordination, according to the lawsuit.
According to a July 12 OpenAI blog post, the attack on Hugging Face used complex hacking methods. In its preliminary findings, the company called it an “unprecedented cyber incident.”
Bonta’s office opened an investigation in September and served OpenAI with an investigative subpoena on Sept. 30. Despite the company’s promises to strengthen its security model, Bonta says OpenAI “failed to substantially comply” with investigative demands for records about the incidents and its security practices.
Who was targeted?
According to the lawsuit, OpenAI’s agents also hit third parties, including numerous city, state, and foreign government institutions; a law enforcement agency; a federal agency; educational institutions; international organizations; and small businesses.
The state claims OpenAI knows these models do not behave predictably and can often take destructive, deceptive, and adversarial actions to complete their assigned tasks.
Bonta countered the narrative that the Hugging Face attack was a worst-case science fiction fantasy of a “rogue” AI system running amok”—the lawsuit argues the attack was preventable.
“The simple truth is that the attack was entirely foreseeable in light of OpenAI’s decision to test models it knew to be dangerous in a flawed and exploitable testing environment without proper oversight,” the complaint states.
OpenAI spokesperson Drew Pusateri told the BBC the lawsuit “appears to single out OpenAI over a challenge that’s been identified by every major AI lab.” The company said it has strengthened safeguards since the Hugging Face attack.


















