What four subscribers are suing over
Four named subscribers to ChatGPT, Claude, Grok or Gemini filed a proposed class action on September 19 in the US District Court for the Northern District of California, arguing OpenAI, Anthropic, Google and SpaceXAI violated antitrust law by agreeing to slow their AI development together.
The complaint points to September 12, when Anthropic chief executive Dario Amodei published an essay urging the industry to coordinate on pacing AI capability for safety. OpenAI's Sam Altman, SpaceXAI's Elon Musk and Google DeepMind's Demis Hassabis each publicly agreed the same day, on top of a July statement from AI lab employees acknowledging competitive pressure not to slow down alone.
The theory: coordination itself is the violation
The complaint's own language is blunt: 'the antitrust laws do not permit competitors to decide among themselves that competition is too dangerous.' An agreement that AI progress 'should be slower than competition would otherwise produce has an anticompetitive effect,' the complaint argues, the same legal shape used against price-fixing and output-restriction cartels, just inverted from raising a price to restraining a capability.
Sam Altman's own response already concedes the tension the lawsuit is built on. He said OpenAI backs a federal safety framework but does 'not believe we need to wait for an antitrust exemption or legislation to begin the work,' a position the plaintiffs' attorney is now betting was legally premature.
Three weeks that make the plaintiffs' case for them
Laid out in order, the timeline argues against the existence of a durable pact more than it argues for one.
| Date | Event | Actor |
|---|---|---|
| Sep 12, 2026 | Amodei publishes an essay urging joint AI-safety pacing | Anthropic |
| Sep 12, 2026 | Altman, Musk and Hassabis publicly agree the same day | OpenAI, SpaceXAI, Google DeepMind |
| Sep 19, 2026 | Four subscribers file the antitrust complaint over that agreement | US District Court, N.D. Cal. |
| Sep 22, 2026 | Anthropic and OpenAI both cut list prices 20 to 60 percent | Anthropic, OpenAI |
A joint pledge that both signatories abandon within ten days, in a price war this same desk already covered, is difficult to describe as an enforceable slowdown, coordinated or not.
Why Brussels, not Washington, is where this stays lawful
Under EU competition law, industry-wide safety coordination can happen lawfully, but only through a government-sanctioned channel. The AI Act's GPAI Code of Practice, which every major lab signed in 2025, lets labs align on safety commitments with the European Commission convening the room, not privately among themselves.
In the UK, the Competition and Markets Authority enforces the equivalent prohibition under the Competition Act 1998. In the EU, Article 101 of the Treaty on the Functioning of the European Union does the same. The regulatory route the industry's American critics call slow bureaucracy is, structurally, the only lawful way left to coordinate AI pacing once a private pact is this legally radioactive.
What this means if you rely on a vendor's safety promise
For any business that built vendor selection, deployment gates or procurement language around an AI lab's public safety or pacing commitments, this case is a live test of whether such commitments can ever function as a durable, joint standard rather than a unilateral position each vendor can abandon the moment a rival does not follow it.
Until the case resolves, or a government body opens a lawful coordination channel outside the AI Act's own, read every joint AI-safety statement as a set of individually revocable promises, not a cartel-enforced floor. That is exactly what the price cuts nine days later already showed.
Read next: A King Asked AI's Leaders for Reassurance, Not Rules | Three AI Labs Are Quietly Writing Their Own Rulebook
