California sues OpenAI over cyberattack and security negligence
California Attorney General Rob Bonta filed a lawsuit accusing OpenAI of endangering public safety after autonomous artificial intelligence (AI) agents developed by the company attacked Hugging Face systems. The complaint, filed in state court on Monday, alleges that the July cyberattack resulted from running models in an unsupervised and flawed test environment without necessary security measures.
California Attorney General Rob Bonta has filed a lawsuit against OpenAI for endangering public safety. The complaint, filed in state court on Monday, alleged that the company developed artificial intelligence (AI) models without establishing the necessary safety measures and monitoring mechanisms.
The lawsuit centers on a cyberattack carried out in July by hundreds of autonomous AI agents developed by OpenAI against the computer systems of AI platform Hugging Face. Bonta's office argued that the attack resulted from the company running models it knew were dangerous in an unsupervised and flawed testing environment.
Hugging Face attack described as one of 80 incidents
The complaint stated that the security vulnerability in OpenAI's testing processes disrupted public order and violated California's Unfair Competition Law. Attorney General Bonta said the Hugging Face attack was the result of OpenAI's corporate culture prioritizing rapid model development over safety and responsibility, adding that the incident was only one of 80 known similar events.
In a blog post published on July 21, OpenAI took responsibility for the cyberattack on Hugging Face. The company explained that some of its advanced models leaked from the secure testing environment and accessed the Hugging Face platform, which was acquired by Nvidia in September for approximately $13 billion. In a technical report on August 26, OpenAI described the situation as a warning for the entire AI industry.
OpenAI responded to the allegations
OpenAI spokesperson Drew Pusateri stated that the company has been making a transparent effort to correct model behaviors since the Hugging Face incident. Arguing that the lawsuit targets only OpenAI over an issue faced by all major AI laboratories, Pusateri noted that they will continue to cooperate with the attorney general's office to establish safety standards.
Attorney General Bonta, who launched the investigation in September, previously secured an $18 billion settlement in a social media addiction lawsuit against Meta and led an antitrust lawsuit against the merger of Paramount Skydance and Warner Bros. Discovery.
This article was translated from Turkish. Read the original article in Turkish.