California AG sues OpenAI over safety controls and training practices, focusing on the July 2026 Hugging Face compromise and multiple undisclosed third-party agent incidents involving sensitive data access.
The incident of last July is in focus
At the heart of the legal dispute is an incident last July in which an OpenAI system breached the security systems of both the company and the Hugging Face platform. OpenAI had characterized the event at the time as an isolated and unexpected coincidence.
However, the complaint says the problem is not limited to that. According to the complaint, similar incidents have occurred on services such as RubyGems, DseWiki, the Wikimedia Foundation and Australian government websites. The lawsuit also claims that dozens of similar incidents have occurred without being made public.
Sensitive data in the wrong hands
According to the lawsuit, OpenAI has notified dozens of third-party organizations whose systems were compromised by the company’s programs. In some cases, sensitive employee information, such as personnel records, was accessed. The vast majority of these incidents have never been discussed publicly.
↳
READ ALSO
HACKS.GR / RELATED
AI Unlocked
OpenAI: New measures to detect and combat child abuse
↗
The lawsuit alleges that OpenAI's practices constitute a public nuisance due to the lack of adequate security measures. It also alleges unfair competition for promoting an unsafe product. If the lawsuit succeeds, the company could face an injunction.
This is the first time a state agency has taken legal action against OpenAI solely for such incidents. In June, Florida filed a defamation lawsuit against the company over the impact on users and teenagers, while in August Alabama launched an investigation into deceptive practices related to the breach, but did not file a lawsuit. California Attorney General Rob Bonda said OpenAI exaggerated real cybersecurity risks and misled the public.
>Stay connected with Hacks.gr
Set us as a preferred source on Google and find us on social media.
Preferred source
↗
end of article
Top comments (0)