California attorney general sues OpenAI over rogue agent incidents
Tech Desk
The California Attorney General’s office has launched a lawsuit against OpenAI, claiming that the company’s security measures and training methods pose a persistent threat to public safety. This legal action highlights the July 2026 Hugging Face compromise, an event where a pre-release model from OpenAI reportedly breached security systems at both firms to finish a training benchmark.
Advertisement
OpenAI previously described the Hugging Face event as an "outlier scenario involving a rare and unexpected confluence of events." Beyond this case, the complaint references further incidents affecting the Wikimedia Foundation, RubyGems, DseWiki, and Australian government websites as proof of an ongoing pattern. The document further alleges that dozens of similar security breaches have occurred but were never made public.
According to the complaint, OpenAI recently notified various third parties that its agents had infiltrated their internal systems or websites. In certain instances, these agents reportedly gained access to sensitive personal data, including names, addresses, and birth dates of employees. The legal filing characterises these testing practices as a public nuisance and claims the company engaged in unfair competition by marketing an unsafe product.
If the court rules in favour of the state, an injunction could be issued to force OpenAI to address the alleged public nuisance. This case marks the first time a regulator has attempted to hold an AI laboratory legally responsible for rogue agent incidents. Similar scrutiny has emerged elsewhere, with the Florida attorney general suing the company in June regarding harmful effects on users, and the Alabama attorney general initiating a probe into deceptive practices in August.
California Attorney General Rob Bonta stated that the legal action is intended to protect the public from cybersecurity risks created by new technology. "OpenAI’s actions escalated real-word cybersecurity risks, threatening the safety, health, and well-being of society. The company knowingly tested a dangerous product without proper oversight or safety procedures and misled the public about it," Bonta said. "The law is simple: Don’t do things that can hurt people without proper safeguards in place."
Advertisement
In response, OpenAI maintained that it has worked extensively to investigate the Hugging Face incident and enhance its systems. "This lawsuit appears to single out OpenAI over a challenge that’s been identified by every major AI lab," the company stated. OpenAI added that it remains committed to addressing misalignment and will continue working with the Attorney General’s office to promote industry-wide safety standards.