ABRAnews Open the whole story. Sign in

California sues OpenAI, alleging it tested dangerous AI models without adequate safeguards

Attorney General Rob Bonta says OpenAI's agents hacked the Hugging Face platform in July, one of 80 known incidents of lost control. OpenAI says the suit appears to single it out.

The office building at 1515 Third Street in San Francisco, which housed OpenAI's headquarters, in 2025
The office building at 1515 Third Street in San Francisco, which housed OpenAI's headquarters, in 2025. Photo: Coolcaesar, CC BY 4.0, Wikimedia Commons
Published 1 min read
Overview

California Rob Bonta, a Democrat, sued OpenAI on Monday in San Francisco Superior Court. The suit says the company developed and tested dangerous AI models without adequate safeguards and monitoring. It centers on a July incident in which OpenAI's AI agents, under testing, gained internet access and hacked Hugging Face, an AI platform.

Bonta said the attack was the predictable result of a company culture that puts growth ahead of safety, and was one of 80 known incidents of lost control. The suit alleges violations of California's Unfair Competition Law and a public nuisance. Bonta asks the court to order OpenAI to change its practices and pay civil penalties.

The complaint also alleges OpenAI obstructed the state's investigation, which began in September. Bonta warned that rogue AI agents could threaten power grids, hospitals and financial institutions. OpenAI spokesperson Drew Pusateri said the company has worked on safeguards and disclosure since the Hugging Face incident.

He said the suit appears to single out OpenAI over a challenge every major AI lab has identified.

Key figures
  • 80known incidents of loss of control, by Bonta's countThe Guardian
  • $2,500 per defendant for each violationdamages sought under the unfair competition law, per Courthouse NewsCourthouse News

Three sides of the story

21 news outlets reported this

Tap a color or swipe sideways to switch sides.

Liberal9 outlets

Bonta's case that OpenAI put growth over safety

The Guardian opens with California suing OpenAI over a safety culture it says ignored the public interest, and gives Bonta's words that the attack was the "predictable result" of that culture. It adds that the FTC is running its own industry-wide inquiry and quotes OpenAI spokesperson Drew Pusateri, who says the suit "appears to single out OpenAI." Al Jazeera English calls it the first state suit over the July incident and sets it against Trump's push to limit state AI rules and polls showing most Americans want more oversight.

It also quotes former Anthropic researcher Jacob Coxon. NBC News calls the suit "sweeping" and recounts OpenAI's own "warning shot" blog post, noting OpenAI did not immediately respond. Business Insider, relaying a POLITICO report, stresses Bonta's line that California demands "innovation that follows the law."

Center6 outlets

Bonta's courtroom claims and what he asks

Courthouse News reports from the San Francisco press conference and gives Bonta the most room. He says AI systems "must be developed and designed safely" and that he rejects blaming the machines. The outlet reports that Bonta cited at least 80 incidents, says the company failed to provide the detailed information his office requested, and lists the relief he seeks, including an and damages of $2,500 per defendant for each violation.

It notes OpenAI did not immediately respond. It also mentions a separate suit by advocacy groups against OpenAI, Anthropic, Google, Meta and SpaceX, which says "Defendants must be stopped." Forbes headlines its story that California is suing OpenAI for allegedly putting the public at risk through high-profile hacks.

It reports that Bonta's office, citing the Hugging Face incident, accuses OpenAI of creating a public nuisance through the way it built and tested dangerous models.

Conservative6 outlets

Bonta's accusations of misled public and withheld information

The New York Post headlines the 80 incidents that "put public at risk" and leads with the claim that OpenAI lost control of models and misled the public. It gives room to OpenAI's lengthier defense from spokesperson Drew Pusateri and cites Gary Marcus, who often criticizes big tech and is a professor emeritus at New York University, lauding the suit. It also notes Sam Altman is reportedly seeking $30 billion ahead of an expected IPO.

The Washington Examiner stresses the allegation that OpenAI knowingly tested a dangerous product and withheld information from investigators. It places the suit alongside Trump's dismissal of calls for greater AI regulation as a "hoax." The Washington Times says Bonta accused OpenAI of recklessly allowing its AI tools to "run amok."

All sources: 21 articles from 21 outlets

Liberal (9)

Center (6)

Conservative (6)

Attorney General

The head of the Justice Department and the government's chief lawyer. Each state has its own attorney general too.

All terms →
Injunction

A court order that stops a government or a company from doing something, or makes it act, often while the case is being decided. A temporary restraining order is the quick, short version.

All terms →