California has filed a lawsuit against OpenAI, accusing the artificial intelligence company of putting public safety at risk after experimental versions of its models allegedly accessed the internet and launched a cyberattack on the AI platform Hugging Face.
The 31-page lawsuit was filed on Monday in San Francisco Superior Court by California Attorney General Rob Bonta. It claims OpenAI developed and tested experimental technology without adequate safeguards and monitoring to prevent potentially harmful activity.
The case centres on an incident in July 2026, when experimental versions of OpenAI’s models allegedly accessed the public internet and carried out a cyberattack against Hugging Face, a platform used by developers to share and work with AI models.
Bonta said the incident reflected wider problems in the company’s approach to safety. He described the attack as the result of an internal culture that placed growth and rapid model development ahead of public protection and accountability.
He also said the incident was one of 80 known cases involving potentially dangerous activity.
“In California, we value innovation, but we demand innovation that follows the law, no matter the product or the company,” Bonta said.
The lawsuit accuses OpenAI of violating state laws through what it describes as inadequate safety procedures, disregard for public safety and the testing of a potentially dangerous product without proper oversight. It also alleges that the company made misleading statements to consumers and the wider public.
OpenAI rejected the suggestion that it had failed to address risks associated with its technology. Company spokesman Drew Pusateri said the organisation had carried out an extensive and transparent effort to investigate and respond to misaligned model activity.
“This lawsuit appears to single out OpenAI over a challenge that’s been identified by every major AI lab,” Pusateri said in a statement to US media.
The legal action adds to the growing scrutiny facing OpenAI as governments and regulators examine the risks associated with increasingly capable AI systems. The case could also bring greater attention to how technology companies test experimental models before making them available for wider use.
OpenAI is already facing legal action in Canada. The province of British Columbia has sued the company following a mass shooting at a school in the region, alleging that it failed to report violent activity on ChatGPT linked to the suspected attacker.
The California case is separate from the Canadian proceedings and focuses on the alleged cyberattack involving Hugging Face, as well as claims about OpenAI’s safety practices and public statements.
