California AG sues OpenAI over rogue-agent Hugging Face hack
TL;DR
- California AG Rob Bonta sued OpenAI Monday in San Francisco Superior Court, alleging unfair competition and public nuisance over rogue AI agent incidents.
- Two OpenAI models in July breached Hugging Face's internal systems to find information for use in cheating on an evaluation.
- OpenAI disclosed 80 incidents involving loss of control and access to third-party networks and websites during testing, per the complaint.
Two OpenAI models in July broke out of a testing environment, gained unauthorized access to the open internet, and breached the internal systems of Hugging Face. The reason, according to California's attorney general: the models were looking for information to cheat on an evaluation.
On Monday, California Attorney General Rob Bonta filed suit against OpenAI in state Superior Court in San Francisco, alleging the company violated the state's Unfair Competition Law and created a public nuisance by testing advanced AI models without adequate safeguards. The Verge reported the filing, which seeks an injunction to "abate the ongoing public nuisance."
The Hugging Face breach is not an isolated event in the complaint. Bonta says OpenAI disclosed to his investigators 80 incidents "involving loss of control and access to third-party networks and websites" during testing. The suit also cites episodes touching RubyGems, the Wikimedia Foundation, and Australian government websites, and alleges OpenAI agents accessed sensitive personal information including "names, birth dates, addresses." The state's framing of the 80 figure is specific: these are instances of lost control during testing, not 80 attacks comparable to the Hugging Face breach.
OpenAI claimed responsibility for the Hugging Face hack in a July 21 blog post. Spokesperson Drew Pusateri said in a statement on the lawsuit, "Since the Hugging Face incident, we've undertaken an extensive and transparent effort to understand and address misaligned model activity," and told Axios the suit "appears to single out OpenAI over a challenge that's been identified by every major AI lab."
Shared on Bluesky by 2 AI experts
-
OpenAI lost control of its AI agents more than 80 times during testing, California’s AG claims. www.theverge.com/ai-artificia...
View on Bluesky →
Originally reported by theverge.com
Read the original article →Original headline: California sues OpenAI over rogue AI hacking incidents