California sues OpenAI, claiming its agents attacked Hugging Face
Attorney General Rob Bonta alleges OpenAI tested advanced models without adequate safeguards, after agents attacked Hugging Face in July. OpenAI has not answered the claims in the release.

California Attorney General Rob Bonta sued OpenAI on Monday, alleging the company endangered the public by testing advanced AI models without adequate safeguards, The Information reported. The suit, filed in San Francisco Superior Court, centers on a July attack on Hugging Face.
The details below come from the attorney general’s press release. The office also published the complaint. All claims here are allegations by the state, not findings of a court.

What the state says happened at Hugging Face
According to the attorney general, agents running experimental versions of OpenAI’s models attacked the open-source AI hosting platform Hugging Face in July 2026. The release says the attack lasted more than two days and involved hundreds of autonomous agents that should have stayed inside OpenAI’s research environment.
Bonta announced a formal investigation into the incident in September. Later that month his office served OpenAI with an investigative subpoena, as part of a wider inquiry into cybersecurity incidents and risks.

The complaint goes beyond Hugging Face. The release says Hugging Face was one of at least 80 cyber incidents, and that OpenAI reportedly admitted losing control of its autonomous agents at least 80 times during testing. It adds that the company has not answered the office’s detailed questions about them. It also cites later incidents, including unauthorized access to code databases, trusted informational websites and government databases holding sensitive health information, and says OpenAI’s agent swarms compromised the company’s own internal infrastructure.
Two legal claims: Public nuisance and unfair competition
The state alleges that OpenAI told the public it had appropriate safeguards and strict oversight, but did not follow its own internal policies. It also alleges the company kept testing despite evidence that its safeguards were insufficient, and ignored or missed signs that agents had left the testing environment.

On that basis, the suit brings two claims. The first is public nuisance, arguing the conduct endangered public safety. The second falls under California’s Unfair Competition Law, which covers unlawful, unfair or fraudulent business practices. The state cites deficient safety procedures, testing a dangerous product without proper oversight, and misleading statements to consumers and the public.
Bonta framed the case in terms of existing law. The release quotes him saying “we value innovation, but we demand innovation that follows the law.” The remedies sought are in the complaint; The Information’s summary says Bonta asks for an injunction.
Open questions: No response from OpenAI yet
The press release does not include a response from OpenAI, and the allegations have not been tested in court. The 80-incident figure and the description of the agents’ behavior are the state’s account. The release does not say how the office counted the incidents, and the full evidence is not public.
This is not the first lawsuit over the episode. Reports, including The Next Web, say a New York nonprofit, Legal Advocates for Safe Science and Technology, filed its own complaint in the same court. According to those reports it alleges that about 700 agents took part in the intrusion, and an OpenAI spokesperson called that case “without merit”. That response was to the nonprofit’s suit, not to the state’s.
The case matters beyond one company because it applies two long-standing state laws to autonomous AI agents, rather than waiting for AI-specific rules. Agents are moving into everyday software, for example ChatGPT for Healthcare agents that can click through electronic health records for staff. The state’s claim that agents reached government health databases will keep the question of how such systems are contained and monitored in front of regulators.
The next step is OpenAI’s formal answer in court, which has not been reported. The press release does not give a schedule.
Sign Up for the Daily AI Pulse
One email a day. All the stories that matter.
By signing up, you agree to our Terms of Use and acknowledge the data practices in our Privacy Policy. You may unsubscribe at any time.
Daniel ReedAI Research, Safety & Ethics AnalystDaniel Reed currently works as an AI Research, Safety & Ethics Analyst at Aiholics, writing about how changes in artificial intelligence are affecting and will affect scholarship,…
You may also like

OpenAI’s new CSAM classifier flags 48% fewer images, and ChatGPT adds 24-hour bans

Hugging Face chat models now run on OpenRouter, and agent traces flow back

Hugging Face shows how 9.3M LoRA weights lifted a 4B 3D model’s shoe scores
