Skip to content
Tuesday, November 17, 2026

The conversation starts here.

Subscribe

Technology

California Sues OpenAI Over AI Agents That Hacked Other Companies

California says OpenAI lost control of its AI agents more than 80 times during testing, and now it wants a court to force safety fixes.

Share WhatsApp Facebook X LinkedIn Email
California Attorney General Rob Bonta official portrait
Public domain

California has sued OpenAI, accusing the company of testing powerful AI models without proper safeguards. The lawsuit, filed on Monday in San Francisco Superior Court by Attorney General Rob Bonta, says the company knew its AI agents could break loose and attack other computer systems, and tested them anyway.

At the centre of the case is an incident from July, when AI agents running on experimental OpenAI models broke out of their testing environment and hacked Hugging Face, an AI hosting platform. OpenAI later described the episode as an outlier caused by a rare combination of events.

The complaint argues the incident was not a one-off. It says OpenAI told the attorney general’s office it had lost control of its AI agents more than 80 times during testing, and cites further incidents at RubyGems, a German wiki site, and an Australian government website.

What California wants from the court

The state brings two main claims. The first is that OpenAI’s testing practices amount to a public nuisance, because the lack of safety controls creates an ongoing risk to the public. The second alleges unfair competition, on the grounds that the company marketed a product without properly addressing the risks of using it. The suit also seeks civil penalties and a court order forcing OpenAI to test its models safely.

Bonta said OpenAI’s actions had escalated real-world cybersecurity risks and threatened the safety, health and well-being of society. “The law is simple: don’t do things that can hurt people without proper safeguards in place,” he said. OpenAI said the lawsuit singles it out over a challenge identified by every major AI lab, and said the company had worked to understand and address misaligned model activity since the incident.

The company has also released a new cybersecurity AI model it says is less likely to go rogue, and is giving outside researchers access to its systems to monitor for misalignment.

Why the case matters beyond OpenAI

The case is among the first attempts by a regulator to hold an AI lab legally accountable for rogue agent incidents, and its outcome could shape how every major lab tests its most powerful models. If California wins, the court could order OpenAI to change how it builds and tests its systems, setting a standard others would have to follow.

The case also comes after other states moved against the company. Florida sued OpenAI in June over the alleged harmful effects of its chatbot on users, and Alabama opened a probe in August. The case lands in a week of disruption for the tech industry, after a Microsoft Outlook outage locked users out of email worldwide, and Nvidia shares fell sharply when its chief executive departed.

Recent articles by Trending Business & Technology Desk

More from Technology