Nonprofits sue five AI giants, ask a court to freeze new model development
Four advocacy groups want a California court to force an emergency pause on next-generation AI training at Anthropic, Google, Meta, OpenAI and SpaceX.
What happened: Four nonprofits — the Irreplaceable Education Fund, Public Citizen, Heat Initiative and the Alliance for Secure AI — filed a civil suit in San Francisco Superior Court against Anthropic, Google, Meta, OpenAI and SpaceX. They're asking for an emergency court order freezing training and release of next-generation AI models until the companies can prove the technology is under human control. The filing lands right after a sweeping Democratic midterm win that could reshape how Washington regulates Silicon Valley.
Why it matters: Congress has debated AI safety for years without passing a law. This lawsuit is an attempt to get a court to do what lawmakers haven't: impose an enforceable stop on frontier AI development. The plaintiffs say they've had to divert their own resources to deal with near-daily problems caused by advancing AI systems, and argue no other industry would be allowed to ship products it admits could cause catastrophic harm and call that inevitable.
How the legal case works: The suit leans on a California law against unfair competition, arguing the companies' AI tools endanger public health and interfere with people's use of their own computer systems, partly by citing documented AI-driven cyberattacks. It also points to the companies' own public warnings that unregulated AI could eventually act outside human control. The plaintiffs want the court to block deployment of newer models until safeguards are proven, not just the ones already released.
What a freeze would actually require: The groups acknowledge a real freeze needs more than a court order: specified rules on what work must stop, government inspection powers, and penalties for violations, potentially including monitoring of major computing facilities. They're pushing for federal evaluators, strict liability for AI harms, and possibly a new regulatory agency, while letting states set tougher rules. On China, they argue a mutual, verified slowdown via treaty wouldn't hand Beijing an advantage — but that argument runs against the Trump administration's stated view that any US restraint helps China win the AI race.