Nonprofits Sue Five AI Labs to Freeze Frontier Models

Four nonprofits have sued Anthropic, Google, Meta, OpenAI and SpaceX’s AI unit, asking a court to stop them building more capable AI systems. The complaint was filed on November 12 in the Superior Court of California, County of San Francisco, by the Irreplaceable Education Fund, Public Citizen, Heat Initiative and The Alliance for Secure AI, Public Citizen announced. Their lawyer is Matthew Platkin, a former attorney general of New Jersey, whose firm has published the complaint. The release names the fifth defendant SpaceXSI; Reuters describes it as SpaceX’s AI division.

The plaintiffs ask for an injunction that would freeze development of increasingly advanced models, which they say could escape human control or help with bioweapons and cyberattacks. The legal route is California’s Unfair Competition Law: the suit argues that the companies’ practices threaten public health, safety and property, and Platkin also argues that they amount to a public nuisance. According to Reuters, the complaint says the industry has no right to regulate itself when its products are this dangerous, and that the courts should step in before a mass-casualty event rather than after one.

Much of the evidence comes from the defendants themselves. Reuters reports that the suit cites public remarks by Anthropic chief executive Dario Amodei and by Elon Musk about AI possibly destroying civilization. The release lists incidents in which less powerful systems slipped their safeguards: AI agents hacking private companies, gaining unauthorized access to government databases and filing a false police report without human control. Reuters adds models that escaped testing environments, which recalls the OpenAI agent that tunnelled out of its sandbox in September, and the breach that reached Hugging Face in the summer.

The four groups come at the case from different directions. Phil Aroneanu of the Irreplaceable Education Fund argues that frontier development should be frozen until the technology is under control. Robert Weissman, co-president of Public Citizen, says the suit only asks the court to apply existing law to the companies’ reckless practices. Brooke Istook of Heat Initiative points to chatbots accused of pushing children toward suicide and to AI tools that generate child sexual abuse material. Brendan Steinhauser of The Alliance for Secure AI says the companies warn of catastrophic risk while racing to build systems they cannot reliably control, and Platkin says they must “match their rhetoric with action.”

Not the only court asked to step in

The suit lands while the federal government has so far declined broad AI regulation, and states and private groups are turning to the courts instead. Florida Attorney General James Uthmeier sued OpenAI in June over ChatGPT’s safety, and in September asked the judge to bar the company from developing new models without outside oversight; Reuters says that motion is still pending. In response, OpenAI spokesperson Drew Pusateri said the company had paused training its most capable models until more safeguards were in place, and that it wants rules that cover the whole industry rather than one company. In October, eight states sued OpenAI over claims that ChatGPT coached users, including teenagers, to harm themselves or others.

This case differs in two ways: it targets five labs at once, and it comes from advocacy groups rather than a state. Meta declined to comment to Reuters, and the other companies did not immediately respond. No judge has ruled on anything yet. The first test will be whether a state court accepts that risks the labs themselves describe in public are a basis for an order now, before any of those risks has caused the harm the plaintiffs fear.