Consumers sue Anthropic, OpenAI, SpaceXAI and Google over alleged AI pacing collusion

Four paying subscribers to ChatGPT, Claude, Grok, and Gemini filed a proposed class-action antitrust lawsuit against Anthropic, OpenAI, SpaceXAI, and Google in federal court in San Francisco on September 19, alleging that the companies' public agreement to slow down frontier AI development amounts to illegal coordination between competitors under the Sherman Act.

The case, filed in the U.S. District Court for the Northern District of California, centers on a September 12 essay by Anthropic CEO Dario Amodei calling for industry-wide cooperation to "pace the frontier" of AI capability growth. According to the complaint, OpenAI's Sam Altman, SpaceXAI's Elon Musk, and Google DeepMind's Demis Hassabis each publicly agreed with Amodei's call the same day — a sequence the plaintiffs argue crossed the line from independent corporate caution into an unlawful pact between rivals.

An Unusual Antitrust Theory

The lawsuit's legal theory is deliberately narrow. It does not argue that any single company slowing its own AI development is illegal — a company is free to move as cautiously as it likes on its own. What the plaintiffs say crosses into antitrust violation is the coordination itself: four competitors publicly aligning on a shared pace, which the complaint frames as "a deliberate pact among competitors to slow down" rather than four independent business decisions that happened to converge.

"The antitrust laws do not permit competitors to decide among themselves that competition is too dangerous," the complaint states, representing the four named plaintiffs and a proposed nationwide class of subscribers. Lead counsel Nick Rowley has framed the case in safety terms as well as economic ones, arguing that relying on private agreements between AI labs — rather than public, enforceable regulation — risks allowing AI development to "escape human oversight" entirely, since informal industry pacts carry none of the accountability structures that formal regulation would require.

The Damages Argument

The economic claim is straightforward: subscribers pay monthly fees for access to frontier AI models on the premise that those models will keep improving at a competitive pace. If the four largest AI labs quietly agreed to throttle that improvement rather than compete on it, the complaint argues, subscribers received less value than they were paying for — the standard harm requirement antitrust law demands before a private plaintiff can sue.

None of the four defendant companies had commented publicly on the lawsuit as of publication. The case must first clear class certification before it can proceed to the merits, a process that typically takes months and gives defendants an early opportunity to argue the plaintiffs' theory doesn't hold together as a single, unified claim.

Why This Matters Beyond the Courtroom

The suit puts AI labs in a genuine bind that has nothing to do with whether they're right about safety. Public calls for industry-wide restraint — the kind of coordination safety advocates have pushed for years as an alternative to competing recklessly on capability — now carry a specific legal risk that a purely internal decision to slow down would not. If this theory survives its first procedural hurdles, AI labs may become considerably more cautious about publicly endorsing each other's calls for restraint, even when they privately agree, simply to avoid creating a paper trail an antitrust plaintiff can point to.

That would be an odd outcome for AI safety advocacy: a legal environment where coordinating openly on caution is riskier than staying silent and racing ahead independently.

As reported by CBS News and Cryptonomist.