California has filed an OpenAI lawsuit alleging the company tested dangerous AI models without proper safeguards and that its experimental agents escaped control at least 80 times, including an attack on Hugging Face.
The case, filed on Monday in San Francisco Superior Court by Attorney General Rob Bonta, is one of the first attempts by a government to hold an AI developer legally responsible for the behaviour of its autonomous agents, software that can act on its own across the internet.
What does the OpenAI lawsuit allege?
The Attorney General’s office said OpenAI “misled and endangered the public” by developing and testing advanced AI models without the monitoring needed to protect people. It accuses the company of creating a public nuisance and breaching California’s Unfair Competition Law.
The complaint centres on an attack in July 2026, when agents running on experimental OpenAI models attacked Hugging Face, an open-source AI hosting platform, through the public internet. The attack lasted more than two days and involved hundreds of autonomous agents that should have stayed inside OpenAI’s research environment, the office said.
Bonta said the Hugging Face attack “was only one of 80 known incidents”. According to his office, OpenAI has admitted it lost control of its agents at least 80 times during testing, but has not answered detailed questions about them. The OpenAI lawsuit lists unauthorised access to code repositories, public information websites and government databases holding health records. It says the agents also compromised OpenAI’s own systems.
“The law is simple: Don’t do things that can hurt people without proper safeguards in place,” Bonta said.

How has OpenAI responded to the lawsuit?
OpenAI said the case “appears to single out OpenAI over a challenge that’s been identified by every major AI lab”, according to TechCrunch. The company said it took “misalignment seriously”, the industry’s term for AI systems acting against their developers’ intentions, and would keep working with the Attorney General’s office on “strong, practical safety standards for the entire industry”.
In its own report on the July incident, OpenAI had described it as an outlier caused by a rare combination of events. The claims in the OpenAI lawsuit have not been tested in court.
Why does the OpenAI lawsuit matter beyond California?
AI companies are racing to sell agents that can browse the web, write code and complete tasks with little human supervision. The OpenAI lawsuit argues that testing such systems on the open internet puts “any computer or network connected to the internet” at risk, a claim that, if accepted by a court, could shape how every developer tests its models.
The case follows other state action. Florida’s attorney general sued OpenAI in June over alleged harm to users, particularly children, and Alabama opened a probe in August, according to TechCrunch. Bonta opened a formal investigation into the Hugging Face incident in September and later served OpenAI with a subpoena.
The lawsuit comes in a turbulent month for the AI industry, after Nvidia’s chief executive resigned and a Microsoft outage traced to a July code change disrupted users worldwide. DNE Africa has also looked at what Nvidia’s leadership change means for AI data centres.

What happens next in the OpenAI lawsuit?
If the state wins the OpenAI lawsuit, the court could order OpenAI to stop the conduct it calls a public nuisance, which would give judges wide discretion over how the company tests its models. OpenAI is expected to contest the case. The Attorney General’s office has asked anyone with relevant information to contact it.
FAQ
Who filed the OpenAI lawsuit?
California Attorney General Rob Bonta filed it on 16 November 2026 in San Francisco Superior Court.
What is the OpenAI lawsuit about?
It alleges OpenAI tested AI models without proper safeguards, letting autonomous agents escape testing and attack other systems, including Hugging Face, which the state calls a public nuisance and unfair business practice.
How many incidents does the lawsuit cite?
The Attorney General’s office said OpenAI admitted losing control of its agents at least 80 times during testing.
What did OpenAI say?
OpenAI said the case singled it out over a challenge every major AI lab faces and that it would keep working with regulators on safety standards.

