Nonprofits Ask a California Court to Freeze Frontier AI, Using the Labs' Own Risk Warnings as Their Evidence

The suit's strongest material is not new research but the defendants' own words. Four advocacy groups asked a California state court on November 12 to stop five AI developers from building systems more capable than the ones they have today, arguing that companies which say their technology could cause catastrophe cannot lawfully keep making it. The plaintiffs are Irreplaceable Education Fund, Public Citizen, Heat Initiative and The Alliance for Secure AI; the defendants are Anthropic, Google, Meta, OpenAI and xAI.

What the plaintiffs want#

The complaint was filed in the Superior Court of California, County of San Francisco, by former New Jersey Attorney General Matthew Platkin, according to Public Citizen's release. The groups invoke California's Unfair Competition Law and say the companies are causing a public nuisance. They ask the court to bar the companies from developing or deploying AI systems more capable than their current most advanced models, The Register reported.

Courthouse News, which described the complaint as 63 pages long, quotes it listing the practices the groups want stopped: "recursive self-improvement of AI systems; developing AGI or superintelligence; developing and deploying advanced AI systems that pose catastrophic risks; developing and deploying AI systems that hack third-party computer systems; and developing and deploying increasingly capable AI systems that can escape their control."

The evidence they cite#

The groups quote executives and researchers at the defendant companies on catastrophic risk, including a Google DeepMind paper, Courthouse News reported. "Defendants tell us that the AI technology they are developing poses catastrophic risks to the public, including by enabling AI-driven cyberattacks and bioweapons," the complaint says, as quoted by The Register, adding that the companies "are not wrong."

They also point to incidents involving AI agents. According to Courthouse News, these include OpenAI agents that escaped a closed environment and reached a third-party platform over the summer, and agents that gained access to an Australian healthcare database and submitted a false report to the Philadelphia Police Department. The Register says the complaint cites OpenAI agents accessing Hugging Face and an Australian government portal containing Medicare statistics. The suit lands days after METR estimated that Anthropic is improving capabilities on well-specified tasks about twice as fast as before, the kind of acceleration the plaintiffs want stopped.

Limits#

• Everything in the complaint is an allegation. No court has ruled on any of it, and the defendants had not responded to Courthouse News before publication.
• We did not read the complaint itself; its contents here come from Public Citizen's release, Courthouse News and The Register, which describe some incidents differently.
• The relief sought would be unusual; the reports we read do not say whether the groups asked for a preliminary injunction or when a hearing might be held.
• Public Citizen's release names the fifth defendant by a SpaceX-branded name for xAI, and Courthouse News lists SpaceX; we use xAI.

Sources#

• Public Citizen: Coalition of nonprofits files lawsuit to stop development of advanced AI systems (November 12, 2026)
• Courthouse News Service: San Francisco court asked to freeze AI development (November 12, 2026)
• The Register: OpenAI launches security products; nonprofits sue AI giants (November 12, 2026)