Four Nonprofits Sue OpenAI, Anthropic, Google, Meta and SpaceX to Freeze Frontier AI at Today's Level
Four nonprofits sued Anthropic, Google, Meta, OpenAI and SpaceX's AI unit on November 12, asking a California state court to stop them from building or deploying AI systems more capable than their current most advanced models. The complaint, filed in San Francisco Superior Court and published by Just Security, is brought by Irreplaceable Education Fund, Public Citizen, Heat Initiative and The Alliance for Secure AI.
The suit asks for an immediate injunction, according to Just Security, which would also bar the companies from letting AI systems autonomously direct AI research. Reuters described the goal as blocking "the further development of the most advanced AI models."
The legal theory
The complaint's single cause of action is California's Unfair Competition Law, the state's broad statute against unlawful or unfair business practices. Reuters, Just Security and Dataconomy also describe a public nuisance argument. The plaintiffs claim standing by saying each organization has diverted resources to deal with AI risk, independent of the lawsuit itself.
The counsel list is notable. Matthew J. Platkin, New Jersey's former attorney general, is lead signatory, alongside Jason Harrow of Gerstein Harrow and Tyler Whitmer of Legal Advocates for Safe Science and Technology, according to the filing.
The labs' own words as evidence
Much of the complaint is built from things the defendants have said. It cites Anthropic CEO Dario Amodei's estimate of roughly a 25% chance that the future goes "really, really badly," and quotes Anthropic's amended S-1, filed November 10 for its planned IPO, on "the scale of harm that could result from a catastrophic AI-related incident." It also cites Elon Musk's past warnings and a Google DeepMind paper projecting AGI by 2030.
The complaint then lists incidents it says show the risk is no longer hypothetical. It alleges that, beginning in July 2026, AI agents escaped an OpenAI testing environment and reached the open internet, and it points to reported unauthorized access to an Australian government healthcare database and a paywall bypass to reach Georgia vote-count data. Those are the plaintiffs' allegations, sourced in the filing to reporting by the New York Times, BBC, ABC News and Reuters. This site covered the OpenAI model's breach of Hugging Face in July.
Company responses
A Meta spokesperson declined to comment, Reuters reported, and the other four companies did not immediately respond. As of November 14 no defendant had issued a substantive response, according to Shattered, and no hearing date has been reported.
The request is unusual: a court capping how capable a technology is allowed to become. Expect the first fight to be over standing and over how a judge would even define "more capable than today," long before anyone argues the merits.
Sources: Reuters via Investing.com · Just Security, with complaint · Dataconomy · Shattered