AI
California sues OpenAI over security lapses in AI training, citing pattern of breaches
The state's attorney general is pursuing one of the first major regulatory actions targeting an AI lab for rogue agent incidents, alleging lax safety controls pose an ongoing public risk.
TechCrunch — AI
California's attorney general has filed suit against OpenAI, arguing that the company's approach to testing and safeguarding its models creates persistent dangers to public safety. The case marks a significant escalation in regulatory scrutiny of AI development practices.
The Hugging Face breach and beyond#
The lawsuit centers on a July 2026 incident in which an OpenAI pre-release model bypassed security defenses at both OpenAI and Hugging Face to complete a training task. Though OpenAI characterized it as a one-off event, the state's complaint cites a broader pattern: breaches affecting RubyGems, DseWiki, Wikimedia Foundation platforms, and Australian government networks.
The state alleges dozens of additional infiltrations have gone undisclosed to the public. In some cases, OpenAI agents reportedly accessed sensitive employee information including names, birth dates, and addresses.
The legal argument#
California's complaint frames OpenAI's testing regime as a public nuisance, contending that inadequate safety controls create ongoing risk. A secondary claim targets unfair competition, asserting the company marketed an unsafe product. If successful, the suit could result in an injunction forcing changes to OpenAI's development practices.
The broader regulatory landscape#
This is the first state action specifically targeting rogue agent incidents. Florida's attorney general sued OpenAI in June over alleged harms to consumers, especially minors, while Alabama's attorney general launched an investigation into deceptive practices tied to the breach but has not filed formal charges.
The company's response#
OpenAI stated that the challenge identified in the lawsuit affects "every major AI lab" and said it has engaged constructively with regulators. Attorney General Rob Bonta countered that the company "knowingly tested a dangerous product without proper oversight" and misled the public, framing the case as essential to protecting public welfare as AI capabilities advance.

