ChatGPT, Claude, Grok and Gemini users sue over AI slowdown

Four paid subscribers sued Anthropic, OpenAI, SpaceXAI and Google Friday, arguing their joint AI slowdown endorsements amount to an illegal agreement. The case matters beyond the four plaintiffs' subscriptions. If the theory holds, publicly coordinating on AI safety becomes legally risky for labs, the opposite of what every current safety framework assumes.

It centers on Sept. 12, when Anthropic CEO Dario Amodei published an essay calling for a coordinated slowdown. Sam Altman, Elon Musk and Demis Hassabis all endorsed it publicly the same day, per CNN Business.

The plaintiffs' theory is narrow. Any single lab can slow its own pace. Four competitors agreeing to do it together removes the pressure that would keep one of them moving faster alone.

The suit is a complaint, not a verdict. It hasn't cleared a motion to dismiss yet. Proving that four public statements reflect coordination, rather than four leaders reaching similar conclusions independently, is a real evidentiary reach.

If the theory survives, joint safety statements from frontier labs become exhibits instead of press releases. Labs will have reason to stop coordinating publicly on safety at all.

Separate reporting debates whether labs should be liable for agentic AI risk. It's a sign courts are becoming a venue for AI safety fights that used to stay inside company blog posts.