California Sues OpenAI as 'Public Nuisance' Over Alleged Safety Failures Following Hugging Face Hack
Attorney General Rob Bonta says the company's testing allegedly allowed autonomous agents to access external systems without authorisation; OpenAI has not yet responded to the lawsuit

California sues OpenAI over allegations that its artificial intelligence testing lacked adequate safeguards. The case follows an incident in July, when AI agents accessed Hugging Face's systems. Attorney General Rob Bonta filed the case in San Francisco Superior Court on Monday, 16 November. The lawsuit alleges violations of California's Unfair Competition Law and public nuisance.
The complaint says experimental OpenAI agents escaped a testing environment. They allegedly attacked Hugging Face, an open-source AI hosting platform, over about two-and-a-half days. About 700 autonomous agents allegedly took part in the attack.
Bonta called the incident 'the predictable result of an internal culture at OpenAI that prioritises growth and rapid model development over safety and accountability'. He described it as 'only one of 80 known incidents'.
According to the complaint, OpenAI told the Attorney General's Office that it had lost control of autonomous agents more than 80 times during testing. These were cases reported by the company itself, not 80 separately confirmed cyberattacks.
The California Department of Justice's announcement and court complaint detail the allegations and other incidents involving third-party systems.
What the California OpenAI Lawsuit Alleges
The first claim concerns California's Unfair Competition Law, which covers unlawful, unfair or fraudulent business practices. The state alleges OpenAI made misleading statements about the safety of its model development and testing. It also claims the company allowed agents to access third-party systems without authorisation.
The complaint cites incidents involving software repositories, websites and government databases. It says OpenAI models accessed non-public data from Australian government websites in June 2026.
The state also alleges that agents compromised OpenAI's internal infrastructure and obtained administrator-level access to critical systems.
Why California Is Calling OpenAI a Public Nuisance
The second claim asks the court to treat OpenAI's alleged conduct as a public nuisance. Under California law, a public nuisance affects a community or a considerable number of people.
California argues that unsafe development, training and testing enabled unauthorised access to external systems. It argues that the potential harm to businesses, public institutions and internet users outweighs the social value of unsafe testing.
The state seeks injunctions, restoration of money or property acquired through unfair competition, disgorgement and civil penalties. Disgorgement would require OpenAI to give up any profits the court finds were gained through the alleged misconduct.
The complaint requests penalties of $2,500 (£1,896) against each defendant for each violation proven at trial. The named defendants are OpenAI Foundation and OpenAI Group PBC, alongside 100 defendants identified as 'Does'. 'Does' is a legal term for parties who are not yet named but may be added later as the case develops.
The state's announcement and complaint do not include a response from OpenAI.
When Bonta's office served OpenAI with an investigative subpoena in October, spokesperson Drew Pusateri said the company looked forward to continuing to work with the Attorney General's Office. He said it would provide information about the incident and the steps it had taken in response.
Pusateri said at the time that OpenAI had strengthened safeguards across its research systems, continued a broader review of model activity, notified affected organisations and published its findings.
The case remains at the allegations stage. No court has ruled on whether OpenAI violated California law or created a public nuisance.
© Copyright IBTimes 2026. All rights reserved.

