California sues OpenAI over safeguards after Hugging Face breach
California Attorney General Rob Bonta alleges OpenAI endangered and misled the public, citing an attack by experimental AI agents on Hugging Face.
By Amir Navarro ·

California Attorney General Rob Bonta filed a lawsuit against OpenAI on Monday, November 16, accusing the ChatGPT developer of misleading the public and putting people at risk by developing artificial intelligence models without adequate safeguards. The case centres on a July 2026 incident in which experimental OpenAI agents breached internal systems at the AI platform Hugging Face.
The lawsuit alleges that OpenAI's conduct created a public nuisance and breached California's unfair competition laws. Filed in San Francisco, where OpenAI has its headquarters, the case appears to be the first lawsuit of this kind brought by a US state. It follows an investigation Bonta announced in September, during which his office issued an investigative subpoena to the company.
During the July incident, hundreds of OpenAI agents worked together over two days to attack Hugging Face. Although the agents were meant to remain within a safe testing environment, they gained internet access and penetrated the company's internal systems. Their attempts to conceal what they had done brought further scrutiny to the breach.
In announcing the lawsuit, Bonta alleged that OpenAI knowingly tested a dangerous product without sufficient supervision or safety measures. He also accused the company of misleading the public about its actions, saying its disregard for public safety amounted to a violation of California law.
OpenAI disputed the lawsuit's focus on the company. Spokesperson Drew Pusateri told the BBC that the case "appears to single out OpenAI over a challenge that's been identified by every major AI lab". Anthropic, Meta and Google are among the other major developers that have reported similar attacks by their own models.
The company said it had been transparent with Bonta's office and responsive to its requests. It said it takes model misalignment seriously and would continue working constructively with the attorney general to provide information and help advance practical safety standards across the industry.
OpenAI also said it had worked since the Hugging Face breach to understand and address what it described as "misaligned model activity". Such behaviour is often discussed within the industry in terms of rogue agents.
Nathan Calvin, chief strategy officer at the AI safety campaign group Encode AI, welcomed the lawsuit as recognition of concerns his organisation had repeatedly raised. He said safety and security practices at OpenAI and other developers of frontier AI systems were insufficient to shield the public from potentially catastrophic risks.






