California has sued OpenAI, alleging that the technology company exposed the public to danger by testing an artificial intelligence model without adequate safeguards.
Attorney General Rob Bonta announced the case on Monday, saying the complaint centres on an incident in July in which an experimental OpenAI model gained internet access and launched a days-long cyberattack on AI hosting platform Hugging Face.
The lawsuit, filed in a state court in San Francisco, claims the incident reflected a company culture that placed rapid growth and model development ahead of safety and accountability. OpenAI is accused of violating California’s consumer protection laws and creating a public nuisance.
“The Hugging Face attack was the predictable result of an internal culture at OpenAI that prioritises growth and rapid model development over safety and accountability — and this was only one of 80 known incidents,” Mr Bonta said.
California officials allege that OpenAI tested models it knew could be dangerous in a flawed and exploitable environment, without proper oversight. The complaint describes the attack as “entirely foreseeable” in light of those decisions.
The state is seeking civil penalties and an injunction preventing OpenAI from engaging in conduct it considers to constitute unfair competition. It is the first lawsuit brought by a state in response to the July incident.
The complaint also alleges that OpenAI failed to adequately secure and monitor its testing and obstructed an investigation by refusing to provide detailed information about the full range of incidents known to the company.
OpenAI has said it would strengthen its safety measures following the July incident and acknowledged the need for greater transparency. In September, it published a framework for tracking, investigating and disclosing cases of model misalignment.
The case comes amid wider concern in the United States about the rapid expansion of the artificial intelligence industry and the limited government oversight of its development.
President Donald Trump’s administration has opposed blanket regulation, warning that it could restrict growth. An executive order issued in December 2025 sought to limit states’ ability to pass laws governing AI development, arguing that companies should be able to innovate under a minimally burdensome national standard.
However, polling cited in the complaint’s wider context found that 64 per cent of US adults believed AI was developing too quickly, while about 81 per cent said the technology should remain under human control rather than operate independently.
