OpenAI, Anthropic, Google and xAI Face Antitrust Lawsuit Over Alleged AI Slowdown Agreement

On Friday, September 18, 2026, a class-action lawsuit (Case No. 3:26-cv-10693) was filed against major AI companies in the U.S. Attorney Nicholas C. Rowley represents the plaintiffs in the 29-page complaint. In the lawsuit, the plaintiffs allege that an agreement among Anthropic, OpenAI, SpaceX AI, and Google to jointly slow the pace of model development is underway and that this agreement restricts the pace of innovation and could therefore violate Section 1 of the Sherman Act. Although no court has found that such an agreement existed, the plaintiffs allege that the agreement took place publicly.

What Does the Lawsuit Allege?

According to the document, the plaintiffs trace the roots of the alleged coordination back to July of this year, with a statement signed by 1,386 employees of frontier AI companies. The statement says that each company and country faces intense competitive pressure not to unilaterally slow the advancement of AI models and calls on the U.S. government to support an international effort to develop tools to control the pace at which AI models advance.

Although this statement shows that concerns about the rapid advancement of AI models and the competitive pressure surrounding a potential slowdown were already being discussed in July, it does not by itself demonstrate that the companies entered into an agreement for this purpose, as the plaintiffs allege. Pacing the Frontier

“Pacing the Frontier” Statement

“We request that the U.S. government support an international effort to develop the technical and governance tools needed to deliberately pace the frontier of automated AI development.”

— 1,386 employees of frontier AI companies. Pacing the Frontier

The plaintiffs also allege that on July 30, Sam Altman, CEO of OpenAI, said he had discussed with White House officials the need to pace the development of AI models as they become more powerful and that OpenAI had helped shape the language of the statement.

“We’ve talked about the need to pace it as the models get more capable, which I think is in everyone’s interest,” Altman told reporters on Capitol Hill. Axios

The Allegation of a “Working Group”

According to the plaintiffs, starting in July, employees below the CEO level at Anthropic, OpenAI, and Google had been meeting regularly to develop industry standards. The plaintiffs also allege that The Information reported on the meetings on September 13 and that OpenAI confirmed on September 15 that it had already been discussing safety issues with Anthropic and Google DeepMind for weeks.

In a September 6 post titled “An Alien Mind,” OpenAI Chief Scientist Jakub Pachocki states that no lab has yet developed sufficient monitoring techniques to keep pace with the rapid development of AI models and that voluntary slowdowns could become commonplace until shared safety standards are established. The plaintiffs use the text as part of their argument that discussions about coordinating a slowdown were already underway before September 12.

The Amodei Factor

On September 11, 2026, one day before Amodei’s essay, Altman was asked during an interview with Fortune about the possibility of developing a common plan among AI executives, including Amodei and Elon Musk. He replied, “I think that will happen,” and declined to “pre-announce private discussions” that he believed should be shared “as a group”. Fortune

According to the plaintiffs, the interview supports their allegation that discussions about a joint plan among frontier AI companies were already taking place. The plaintiffs argue that Amodei’s September 12 essay then served as a public proposal for coordination, including limits on certain aspects of AI development and mechanisms to verify compliance. Dario Amodei — We Must Pace the Frontier

As previously reported here , Amodei did indeed propose a plan to slow the pace of AI development. In his essay, he outlines a three-step plan involving government participation and discusses the antitrust issues that could arise from coordination among competing companies.

“The Public Acceptance”

In the document, the plaintiffs allege that Musk, Altman, and Hassabis’s responses to Amodei’s proposal could constitute an agreement in the same way that private communications could. The plaintiffs highlights Altman’s response — “I agree with Dario that we need to pace the frontier” — and Musk’s — “Dario is right” — as evidence of the alleged public acceptance.

Combined with the statements and events before and after these responses highlighted above, these public statements form part of the narrative the plaintiffs use to support their lawsuit.

If the case proceeds to discovery, the plaintiffs expect to examine evidence including communications between the companies, meeting records, internal development plans, research budgets, compute allocations, training plans, release schedules, and competitive analyses. According to the plaintiffs, this information could help determine whether the companies actually agreed to jointly slow the development of their competing products.

What Still Needs to Be Proven?

In addition to demonstrating that the alleged agreement actually existed, the plaintiffs will also need to establish the harm caused to consumers of the AI models involved. The plaintiffs allege that subscribers have been receiving less improvement for the same price and characterizes this difference as an “overcharge.”

According to the plaintiffs, the alleged agreement is recent and its full effects have not yet appeared in released products. The plaintiffs say damages could be estimated by comparing what actually occurred with a counterfactual scenario without the alleged coordination, using data on investment, compute, training, benchmarks, prices, demand, and consumer behavior.

What Is the Lawsuit Seeking, and Who Could Benefit?

The lawsuit seeks to represent consumers in the United States who, since September 12, 2026, directly purchased individual premium subscriptions to Claude, ChatGPT, Grok, or Gemini. The plaintiffs also leave open the possibility of changing the class period if evidence obtained during the case indicates that the alleged agreement began before that date.

The plaintiffs are also seeking treble damages under Section 4 of the Clayton Act and an injunction to prevent the alleged coordination from continuing.

As of the date of publication, no court has concluded that such an agreement existed or that any of these companies violated antitrust laws.