Anthropic, OpenAI, Google and SpaceXAI are facing a proposed US class action accusing the four AI rivals of illegally coordinating to slow improvements to their models in the name of safety, turning an industry debate over frontier-AI risks into an unusual antitrust case.
The complaint was filed on September 18 in the US District Court for the Northern District of California by four consumers who pay for ChatGPT, Claude, Grok and Gemini. They seek to represent a nationwide class of paid subscribers and accuse the companies of violating Section 1 of the Sherman Antitrust Act, which prohibits certain agreements that restrain competition. The allegations have not been tested in court.
The case centers on events six days earlier. Anthropic CEO Dario Amodei published an essay calling for “industry-wide coordination” to slow the rate at which frontier AI capabilities advance so that safety measures can keep pace. Sam Altman, Elon Musk and Google DeepMind CEO Demis Hassabis subsequently expressed support for elements of his proposal.
The plaintiffs argue those public responses amounted to an agreement among competitors to restrict the rate of product improvement. Their economic theory is unusual but straightforward: subscribers continue paying for AI services that, they allege, would become more capable more quickly under unrestricted competition. The complaint characterizes this as a form of output restriction.
Whether public expressions of agreement are enough to establish an unlawful antitrust pact is now a question for the court. No finding has been made that the companies actually implemented a coordinated slowdown.
The legal tension was already visible before the lawsuit. Amodei had suggested that Washington might need to provide a narrow antitrust waiver for some cross-lab safety discussions. Altman instead supported a federal framework with common safety requirements while arguing that companies did not need to wait for an antitrust exemption to begin cooperating.