Alleged July 2026 Working Group Is Central Evidence in Antitrust Suit Against Anthropic, OpenAI and Google
Four consumers have filed suit against the four leading AI labs, alleging that their public coordination around "pacing the frontier" is in reality an unlawful secret agreement among competitors. The case lands in a week when the labs are simultaneously opening up to federal audits and more than 100 safety evaluators are demanding real access.
The Plaintiffs and the Defendants
On September 18, 2026, a class action was filed in the U.S. District Court for the Northern District of California, San Francisco Division. According to Unite.AI's coverage of the suit, four consumers — Charles Buist and Nick Spetsas of Florida, and Cheyenne Hunt and Christine Bullock of California — are suing both individually and on behalf of a proposed nationwide class. The defendants are Anthropic, OpenAI, SpaceXAI and Google.
The core of the allegation is serious: that the companies agreed among themselves to slow the pace of development of their competing frontier models, in violation of federal antitrust law. It is important to emphasize that these are allegations from a lawsuit, presented through Unite.AI's characterization of the document. None of the defendants has responded in the available source material, and neither the complaint itself nor Amodei's essay is available as primary documents.
The Alleged Timeline
The lawsuit alleges that the coordination began months before the public exchange in September. Representatives of Anthropic, OpenAI and Google are said to have formed a working group in July 2026 that met regularly to work on an industry-wide standards organization, according to the allegations in the document.
Then, on the morning of September 12, 2026, Anthropic CEO Dario Amodei published the essay "We Must Pace the Frontier." In it he writes, as the lawsuit quotes him: "We must slow the pace at which we improve the capabilities of AI models." According to the allegations in the suit, the responses came lightning-fast: Within about an hour, Elon Musk, who founded and controls SpaceXAI's Grok business, is said to have publicly endorsed the proposal; Sam Altman is said to have written that he agreed with Amodei and committed OpenAI to the plan's first step; and Demis Hassabis is said to have approved the essay's direction.
The lawsuit also draws in a WIRED report from September 10, 2026, which reported that OpenAI had sought guidance from Members of Congress on whether a coordinated industry-wide slowdown could conflict with antitrust law, as well as a Fortune interview on September 11 in which Sam Altman said he believed a shared industry plan was going to happen, without going into private discussions.
For the plaintiffs, it is precisely this speed and form — private coordination among competitors, confirmed publicly — that is the evidence they need.
How the Case Is Scoped
At the same time, the plaintiffs are deliberately limiting the case. According to Unite.AI, they expressly challenge none of the companies' unilateral decisions on safety, testing, environmental impact or development pace, nor do they challenge the companies' advocacy toward Congress or the White House. The case concerns only alleged coordination among rivals, not the safety thinking itself.
The Same Week: The FRONTIER Act and the Evaluators' Letter
The lawsuit landed in a week marked by three simultaneous threads in the same governance debate. On September 16, OpenAI confirmed that it supports parts of the bipartisan FRONTIER Act proposal in Congress, which would require the largest AI companies to submit their language models to independent audits. It is the first time OpenAI has backed a federal third-party audit mandate, according to CBS News, which cites the company's chief of global affairs, Chris Lehane. But OpenAI has not endorsed the proposal in its entirety.
On the same day the lawsuit was filed, more than 100 AI experts and evaluators published an open letter organized by the AI Evaluator Forum, with Geoffrey Hinton among the signatories. The letter warns that evaluators lack the necessary resources and protections to test the safety of frontier models — a pressure that comes right after Amodei's weekend proposal for "employee-like access." According to CNBC, Sam Altman, Elon Musk and Microsoft CEO Satya Nadella have publicly supported the proposal, but have yet to take a position on the practical questions: Which evaluators will be selected, and how deeply will they be allowed to inspect closely guarded technologies?
Conrad Stosz of the AI Evaluator Forum described the letter as part of an effort to hold frontier labs accountable to their recent promises of more thorough third-party testing, while acknowledging that the companies could ignore it.
What We Don't Know
Several caveats are necessary. Everything about the alleged pact consists of allegations known through Unite.AI's coverage of the lawsuit — not established facts, and not confirmed by the companies. No defendant statements appear in the available material. There is also a chronological tension the sources do not resolve: OpenAI simultaneously supports federal audit requirements and is being sued for an alleged secret slowdown, and the evaluator letter both welcomes and pressures Amodei's access proposal.
What the lawsuit puts squarely to the test, however, is concrete: Where is the line between voluntary safety coordination and an unlawful agreement to restrain competition? The judge in San Francisco will have to decide.