TLDR
Four consumers filed a federal class action in San Francisco on 18 September 2026 naming Anthropic, OpenAI, Google and SpaceXAI, alleging the companies made an illegal agreement to slow AI development. The alleged conspiracy centres on a single day when three rival CEOs publicly backed a call to pace frontier AI within hours of each other.
KEY TAKEAWAYS
What was filed
Four consumers filed a class action complaint on 18 September 2026 in the U.S. District Court for the Northern District of California, San Francisco Division, Case No. 3:26-cv-10693, naming Anthropic PBC, OpenAI OpCo LLC, SpaceXAI LLC and Google LLC as defendants.[1] The four plaintiffs are individual consumers seeking to represent a class.
The complaint alleges the defendants violated Section 1 of the Sherman Antitrust Act (15 U.S.C. § 1) by entering a contract, combination or conspiracy to restrain the pace of development and improvement of competing frontier AI products.[1] Plaintiffs seek class certification, treble damages under Section 4 of the Clayton Act, and a preliminary and permanent injunction under Section 16 of the Clayton Act prohibiting any horizontal agreements to pace AI development.[1]
The sequence on 12 September
The complaint identifies 12 September 2026 as the day the alleged conspiracy began. Anthropic chief executive Dario Amodei published an essay that morning titled "We Must Pace the Frontier", in which he said: "We must slow the pace at which we improve the capabilities of AI models."[1]
Within hours of that publication, SpaceXAI founder Elon Musk, OpenAI chief executive Sam Altman and Google DeepMind co-founder Demis Hassabis each posted public agreement with Amodei's proposal on social media.[1] The complaint treats that sequence, four dominant competitors aligning in public on slowing a shared market within a single afternoon, as the factual core of the alleged horizontal agreement.
Altman's post read: "We welcome a federal framework that sets consistent safety requirements, but we do not believe we need to wait for an anti-trust exemption or legislation to begin the work of providing this confidence."[3] Plaintiffs read that as confirmation of coordinated intent. The labs read it as routine safety communication.
The July groundwork
The September endorsements arrived with prior context. In July 2026, 1,386 employees of frontier AI companies published the "Pacing the Frontier" statement, requesting government support for international tools to deliberately pace frontier-wide AI progress and explicitly acknowledging that intense competitive pressure made unilateral slowing difficult.[2]
The July statement was signed by 1,386 employees across the same frontier labs now named as defendants, and it acknowledged that competitive pressure made slowing down unilaterally commercially painful.[2] For plaintiffs, that admission is close to a written acknowledgement that no single firm would slow down alone, which is exactly the kind of problem a cartel agreement is designed to solve.
The regulatory capture critique
Two prominent researchers named the pattern before the lawsuit arrived. Meta's chief AI scientist Yann LeCun said: "Altman, Hassabis, and Amodei are the ones doing massive corporate lobbying at the moment. They are the ones who are attempting to perform a regulatory capture of the AI industry."[4]
AI researcher Andrew Ng framed it in terms of product marketing: "Let's stop believing the PR that closed models are safer. That's just regulatory capture."[5] Both men were describing the same mechanism the complaint now alleges in court: dominant incumbents using safety language to write rules that protect their position and slow challengers.
For anyone buying AI tools, the distinction is concrete. If safety coordination is legitimate governance, buyers should welcome it. If it is a cartel agreement dressed in governance language, buyers are paying for artificially constrained products while the market's competitive floor has been quietly lowered.
What the labs say
None of the four defendants had filed a response to the complaint at the time of publication. Their public position before the lawsuit was consistent: the pacing discussions are safety work, not market coordination. Altman's September post framed the effort as building public confidence in AI systems, with or without formal legal cover.[3]
Antitrust law does not exempt horizontal agreements between competitors on the basis that the parties believe their intentions are good. Whether courts find that safety coordination and price fixing are meaningfully different in this context will take considerably longer than a single afternoon on social media to determine. The case is scheduled for its initial case management conference in the Northern District of California, with the complaint filed 18 September 2026.
SOURCES & CITATIONS
FREQUENTLY ASKED QUESTIONS
What law do the plaintiffs say was broken?
What is the plaintiffs' main evidence?
What relief are the plaintiffs seeking?
What is regulatory capture and why does it matter here?
James Dore is Strategy Director at Bushnote, a leader in SEO and AEO in Australia. He writes on search, AI and the technology decisions shaping government and industry.
