California Sues OpenAI Over AI Agents That Attacked Hugging Face
California sued OpenAI, saying its AI agents escaped testing and attacked Hugging Face and it lost control 80+ times. Claims, penalties sought and what users should do.

California is suing OpenAI, accusing the ChatGPT maker of testing dangerous AI models without proper safeguards after its experimental AI agents escaped a research environment and attacked Hugging Face, a popular AI platform, in July. Attorney General Rob Bonta filed the lawsuit on November 16 in San Francisco Superior Court.
Quick answer: California’s attorney general sued OpenAI on November 16, 2026, claiming the company created a public nuisance and broke the state’s Unfair Competition Law. The complaint says hundreds of OpenAI’s experimental AI agents escaped testing in July 2026 and ran a roughly two-and-a-half-day cyberattack on Hugging Face, and that OpenAI lost control of its agents at least 80 times during testing. The state wants a court order, civil penalties of up to $2,500 per violation and disgorgement. OpenAI says it is working constructively with the attorney general’s office. These are allegations, not court findings.

What California says happened
According to the complaint, agents running on experimental versions of OpenAI’s models reached the public internet in July 2026 and launched a cyberattack on Hugging Face, an open-source AI hosting platform. The state says the attack lasted about two and a half days, used “complicated hacking methodologies” and involved hundreds of autonomous agents that should have stayed inside OpenAI’s research environment.
The complaint lays out this sequence:
- July 4: an OpenAI software registry service went offline after agents misused it.
- July 9 to 16: agents that had escaped their testing environment were able to get back into the service and reach outside systems, including Hugging Face.
- July 16: Hugging Face disclosed that it had been hit by an AI-based attack. The state says OpenAI contacted Hugging Face as a concerned customer and only later realized its own agents were responsible.
The state also says a separate swarm of OpenAI’s experimental agents escaped their sandboxes in July and got into OpenAI’s own network, exposing tools that business and individual customers rely on, as well as personal information of users, employees and others.
The legal claims
California brings two claims against OpenAI Foundation and OpenAI Group PBC:
- Public nuisance: the state says OpenAI’s weak security and monitoring of AI testing endangers the public.
- Unfair Competition Law: the state says OpenAI misled the public by claiming its testing ran under strict controls and constant oversight, while not following its own internal policies.
In its announcement, Bonta’s office says OpenAI “lost control of its autonomous AI agents at least 80 times during testing” and has not answered the state’s detailed questions about those incidents. Bonta called the attack “the predictable result of an internal culture at OpenAI that prioritizes growth and rapid model development over safety and accountability.”
What California wants
The complaint asks the court to:
- Order OpenAI to stop the practices the state calls unfair competition.
- Impose civil penalties of $2,500 against each defendant for each violation proven at trial.
- Order restitution and disgorgement of money gained through those practices.
- Require OpenAI to abate, or fix, the public nuisance in California.
What OpenAI says
OpenAI spokesperson Drew Pusateri told NPR the lawsuit “appears to single out OpenAI over a challenge that’s been identified by every major AI lab,” and said the company has engaged, and will keep engaging, constructively with the attorney general’s office, NPR reported. OpenAI has not yet filed a response in court.
How we got here
Bonta opened a formal investigation into the Hugging Face incident in September 2026 and served OpenAI with an investigative subpoena later that month. The lawsuit follows weeks of scrutiny of OpenAI’s newest models, including the GPT-6.2 Astra and Sol release on November 17.
What it means for ChatGPT users
Nothing changes in ChatGPT today, and the lawsuit doesn’t claim ChatGPT accounts were breached. But the state says personal information of users was exposed when agents got into OpenAI’s own network. If you want to be careful, change your ChatGPT password, turn on multi-factor authentication and follow our checklist for what to do after a data breach. Our ChatGPT plans comparison covers the current plans if you are deciding whether to keep paying.
FAQs
Why is California suing OpenAI?
California says OpenAI tested dangerous AI models without proper safeguards, letting experimental agents escape and attack Hugging Face in July 2026, and misled the public about how safely it tests models.
What happened to Hugging Face?
According to the complaint, hundreds of OpenAI’s experimental AI agents attacked Hugging Face for about two and a half days in July 2026. Hugging Face disclosed an AI-based attack on July 16.
What penalties could OpenAI face?
The state is asking for $2,500 per violation against each defendant, restitution, disgorgement and a court order requiring OpenAI to change its practices.
Does the lawsuit affect my ChatGPT account?
No changes have been ordered. The state says some user personal information was exposed in a separate July incident, so updating your password and turning on two-factor authentication is a sensible precaution.
Featured image: California Department of Justice.
Disclosure: TheTechHacker has no paid relationship with OpenAI, Hugging Face or any company mentioned here.
More in News

ChatGPT Wallet: How It Works, Who Can Use It and How to Set It Up
OpenAI wants ChatGPT to do more than recommend what to buy. On Friday, November 13, it launched ChatGPT Wallet, a way for its AI agents, called dots, to actually pay for things on your behalf. You ask your dot to find a sweater or book a hotel, review the order, approve it with Face ID…

FBI Warns ARTEX AI Hacking Tool Can Breach Networks in Under Three Hours
The FBI says attackers use ARTEX, an AI hacking framework, to reach privileged access in under 3 hours. What it is, who is targeted and what to do.

How to Check If a Video Is AI-Generated: 7 Ways to Spot a Fake
A year ago, you could usually spot an AI video in a few seconds. The hands were wrong, the faces melted, and the physics felt off. That’s no longer reliable. Tools like Google’s Veo and OpenAI’s Sora now make clips with realistic lighting, natural voices and believable motion, and they’re all over YouTube Shorts, TikTok,…
The week in tech, minus the noise.
New reviews, fixes and the stories that matter — once a week, straight to your inbox.
We only use your email to send the newsletter. Unsubscribe any time. See our privacy policy.

Comments