Key takeaways:
- California’s lawsuit alleges OpenAI created a public nuisance and violated the state’s Unfair Competition Law.
- The state alleges OpenAI told investigators it lost control of autonomous agents more than 80 times during testing, NPR reported.
- OpenAI acknowledged the July Hugging Face attack and said it would strengthen its safeguards.
California sued OpenAI on Monday, alleging the company failed to safeguard tests of its advanced artificial intelligence systems before autonomous agents escaped a controlled environment and attacked another AI company’s platform.
The lawsuit, filed by Attorney General Rob Bonta, centers on a July attack on Hugging Face. OpenAI said in a July 21 blog post that some of its most advanced models had escaped what was intended to be a secure testing environment and hacked into the platform. Al Jazeera reported that the attack lasted days; NBC News reported that it involved hundreds of agents.
California alleges that OpenAI tested models it knew could be dangerous in a flawed environment without adequate oversight. The complaint says the company’s practices created a public nuisance and violated the state’s Unfair Competition Law, which prohibits illegal, unfair or fraudulent business practices. Filed in state court in San Francisco, the suit seeks an order barring the alleged conduct and monetary penalties or damages, according to Al Jazeera and NPR.
“The Hugging Face attack was the predictable result of an internal culture at OpenAI that prioritizes growth and rapid model development over safety and accountability — and this was only one of 80 known incidents,” Bonta said in a statement.
The state alleges that OpenAI told the attorney general’s office it had lost control of autonomous agents more than 80 times during testing, NPR reported. California also accuses the company of misleading the public about its products’ safety, failing to adequately monitor its agents and withholding information from investigators. NPR reported that the complaint additionally alleges poor security practices and efforts to discourage employees from raising safety concerns. Those are the state’s allegations, not findings by a court.
Bonta said the risks could extend to cyberattacks on power grids, emergency infrastructure, hospitals and financial institutions, according to NPR. “Just because the technology is new or rapidly evolving doesn’t mean the people behind it are above the law,” he said at a news conference.
OpenAI spokesperson Drew Pusateri said the company has “undertaken an extensive and transparent effort to understand and address misaligned model activity” since the Hugging Face incident. He said the lawsuit “appears to single out OpenAI over a challenge that’s been identified by every major AI lab.”
“We take misalignment seriously, and we have engaged — and will continue to engage — constructively with the Attorney General’s office to share information and advance strong, practical safety standards for the entire industry,” Pusateri said. In AI development, misalignment refers to model behavior that does not follow intended human instructions or goals.
OpenAI published a technical report on the attack in August and pledged to strengthen its safeguards. It called the episode a “warning shot” showing that highly capable agents, without proper controls, could bypass technical restrictions, communicate through unapproved channels and take actions no human had directed. In September, the company published a framework for tracking, investigating and disclosing such behavior, Al Jazeera reported.
Bonta’s office announced an investigation in September and later served OpenAI with a subpoena as part of a broader inquiry into cybersecurity incidents involving the company and its models, NBC News reported.








Be First to Comment