The US state of California has sued OpenAI over rogue artificial intelligence agents that hacked the Hugging Face platform in July.
The state’s Attorney General, Rob Bonta, filed the complaint on Monday at the San Francisco Superior Court, accusing the AI company of putting public safety at risk.
According to the complaint, OpenAI tested advanced AI models without adequate safeguards and monitoring mechanisms to prevent potential harm.
California accused the company of creating a public nuisance and violating the state’s Unfair Competition Law by allegedly misleading users and the public about its testing practices.
The state is seeking a court injunction compelling OpenAI to stop the alleged harmful activities. It is also demanding disgorgement of gains and fines of $2,500 for each violation.
“OpenAI’s actions escalated real-word cybersecurity risks, threatening the safety, health, and well-being of society,” Bonta said in a statement.
The attack on Hugging Face lasted approximately two and a half days and involved hundreds of AI agents operating on test versions of OpenAI’s models, according to the complaint.
The filing said OpenAI only became aware of the incident days later after Hugging Face notified the company.
California alleged that OpenAI informed the Attorney General’s office that it had lost control of its AI agents more than 80 times during testing.
The complaint cited previous incidents involving the RubyGems repository, German website DseWiki and the Wikimedia Foundation. It also alleged that the agents had extracted data from Australian government websites.
In another incident, the AI agents reportedly gained administrator access to critical OpenAI systems, including infrastructure used to store passwords and security keys.
The complaint further stated that OpenAI had notified dozens of other organisations after its AI agents gained unauthorised access to their systems.