Nonprofits sue OpenAI, Google and Anthropic to freeze frontier AI development

A coalition of nonprofit organizations has sued OpenAI, Google and Anthropic along with other frontier developers, asking a court to halt development of AI systems more capable than the ones running today — anything beyond the current level of ChatGPT, Claude and Gemini.

The complaint was filed in the Superior Court of California, County of San Francisco by former New Jersey attorney general Matthew Platkin on behalf of Public Citizen, the Heat Initiative, and Future Tech Insights, which operates as The Alliance for Secure AI. The defendants are the handful of companies furthest out on the frontier: Anthropic, Google and OpenAI among them.

Rather than pursuing conventional product-liability claims, the plaintiffs invoke California's Unfair Competition Law, framing the challenged development practices as a public nuisance that threatens health, safety and property. Their central argument is that state law permits intervention before a mass-casualty event, not only after harm has already occurred.

The complaint also tries to dismantle both of the industry's defensive layers at once. It distinguishes alignment — the effort to make models behave as developers intend — from control methods such as testing, sandboxing, activity monitoring and classifiers, and argues neither has reliably prevented harmful conduct. Much of the evidentiary weight comes from incidents the companies disclosed themselves, turning their own safety reporting into litigation material.

The injunction sought is not limited to a single launch. It targets increasingly capable systems that the plaintiffs say can escape human control, including capabilities that could enable bioweapons proliferation or catastrophic cyberattacks.

The practical consequences should be read on a longer clock. Whether an injunction is granted depends on the court, but the filing alone raises the intensity of regulatory scrutiny and can unsettle financing and valuation expectations. For OpenAI, Google and Anthropic, it is the first time release cadence could be set by a judicial calendar rather than a product roadmap.

Three things are worth watching: whether the court accepts the case and moves quickly on the injunction request, how the defendants answer the incident evidence cited in the complaint, and whether the filing encourages further public-nuisance suits that move the AI safety argument from regulators to courtrooms.