Four consumers sued Anthropic, OpenAI, SpaceXAI — the entity through which Elon Musk now controls the Grok chatbot business — and Google in federal court in San Francisco on September 18, 2026, alleging the four companies violated the Sherman Act by agreeing among themselves to slow how fast their AI models improve. The class action turns Anthropic CEO Dario Amodei’s September 12 essay calling to “pace the frontier,” and the endorsements from rival CEOs that followed within the hour, into the factual centerpiece of a proposed antitrust case seeking treble damages.
- Filed: September 18, 2026, U.S. District Court for the Northern District of California, San Francisco Division.
- Plaintiffs: Charles Buist and Nick Spetsas of Florida; Cheyenne Hunt and Christine Bullock of California — each suing individually and on behalf of a proposed nationwide class.
- Defendants: Anthropic PBC, OpenAI OpCo LLC, SpaceXAI LLC, and Google LLC.
- Claim: Section 1 of the Sherman Act, pled as an unlawful-per-se restraint and, in the alternative, under quick-look and rule-of-reason analysis.
- Lead counsel: Nicholas C. Rowley, joined by Andrew T. Tutt, R. Stanton Jones and Jakob Z. Norman of Trial Lawyers for Justice.
- Relief sought: Treble damages under the Clayton Act and an injunction against horizontal agreements on AI development pace; jury trial demanded.
- Reported by: Unite.AI (Mira Kellan), September 18, 2026.
What the complaint alleges
The filing’s core theory, quoted directly: “an agreement among rivals to reduce the quality of their products and the rate at which those products improve is an agreement to restrict output,” and that restraint does not become lawful because the products happen to be new. The complaint is careful about what it is not challenging — the plaintiffs say they take AI safety seriously, that guardrails should be set by the public through regulation and juries rather than by the defendants themselves, and that they are not contesting any defendant’s unilateral safety decisions or its advocacy to Congress or the White House. The claim is narrower: that what looks like four companies each independently choosing to slow down was, on the complaint’s telling, coordinated.
The timeline the complaint builds toward that inference: representatives of Anthropic, OpenAI and Google formed a working group in July 2026 to develop an industry standards body, and on July 14 Google DeepMind co-founder Demis Hassabis publicly proposed a U.S.-led standards body modeled in part on FINRA. On September 6, OpenAI chief scientist Jakub Pachocki published an essay, “An Alien Mind,” describing coordinated slowdown as one option available to frontier developers. A September 10 WIRED report said OpenAI had asked members of Congress whether an industry-wide coordinated slowdown would violate antitrust law, and in a September 11 Fortune interview OpenAI CEO Sam Altman said he expected a common industry plan to emerge.
Then, on the morning of September 12, Amodei published We Must Pace the Frontier, writing: “We must slow the pace at which we improve the capabilities of AI models.” The complaint alleges that within about an hour, Musk publicly endorsed the proposal, Altman said he agreed and committed OpenAI to its first step, and Hassabis endorsed it and tied it to the standards body he had proposed in July. On September 14, Altman said AI progress would proceed more slowly and that OpenAI would not wait for an antitrust exemption or legislation before working with other labs. On September 15, OpenAI’s global policy chief, Chris Lehane, confirmed OpenAI had been working with Anthropic and Google DeepMind on the issue for several weeks. The complaint also cites a separate July 2026 statement, “Pacing the Frontier,” carrying 1,386 signatories — including Amodei, Anthropic co-founders Jared Kaplan and Jack Clark, Pachocki, OpenAI chief research officer Mark Chen, and DeepMind’s Shane Legg — which asked governments to support an international effort to pace frontier AI development because of “intense competitive pressure not to unilaterally slow” it.
Amodei’s essay itself proposed three steps: embedded third-party evaluators with employee-like access (a measure Anthropic says it has adopted unilaterally); coordination among frontier developers in democratic countries on common safety standards and limits on the rate of unchecked progress; and, where possible, coordination extending to authoritarian governments. A footnote in the essay concedes the second step depends on government mediation or antitrust waivers, and urges companies to work together voluntarily in the meantime — the exact sentence the complaint now treats as the plan the four defendants allegedly carried out. The proposed class covers U.S. residents who bought a paid individual subscription to ChatGPT, Claude, Grok or Gemini from September 12, 2026 onward, on a theory that a slower rate of improvement is a lower-quality product sold at an unchanged price — the defendants are alleged to hold roughly 80% of that paid-subscription market.
Not the Accenture evaluation deal
This is a different Anthropic story from the one CASRAI covered on September 18 in Anthropic and Accenture Commit $1B Each to Scale AI Model Evaluation. That piece is about a commercial arrangement to embed third-party evaluators inside Anthropic — the same practice Amodei’s essay lists as step one of “pacing the frontier,” and unilateral, not the subject of any antitrust claim. This lawsuit is about step two: whether Amodei’s call for frontier developers to coordinate on the rate of capability improvement, and the rival CEOs’ public endorsements of it, crossed the line from parallel individual conduct into an unlawful horizontal agreement. None of casrai.org’s 17 existing frontier-ai-safety news posts have covered this specific antitrust framing of the pacing pledge before — this is CASRAI’s first.
NIKOLAI’s read: what happens when a pacing pledge becomes a plaintiff’s exhibit
CASRAI’s own NIKOLAI project — an independent, unendorsed reference dictionary for frontier-AI-safety terminology, not a standard adopted or enforced by any lab, regulator, or court — already draws a line through exactly this territory in its Commitments and Governance track (N9), and it is a genuine, verified tie-in rather than a stretched one: NIKOLAI separates two elements that this lawsuit’s theory turns on treating as one thing.
The first is Commitment, defined on its live element page as “a public statement by an identified party (a developer, a government body, or a named individual signatory) to do or not do something, carrying an explicit modality (binding / unilateral / conditional / aspirational), a scope, an effective or target date, and a status value that can be tracked over time.” NIKOLAI’s own shadow-mapping crosswalk for this element already cites Anthropic’s Responsible Scaling Policy, OpenAI’s Preparedness Framework, Google DeepMind’s Frontier Safety Framework, and the Seoul Summit’s Frontier AI Safety Commitments as prior instances of the pattern — each logged as a CASRAI shadow mapping, not a designation any of those organizations has endorsed.
The second is Industry-Wide Recommendation, defined as “a standard a developer or body states all frontier developers should meet, recorded separately from any statement about whether the author itself currently meets that standard.” NIKOLAI’s own crosswalk for this element already logs Anthropic’s prior “recommendations for industry-wide safety” — recommendations Anthropic itself has said it does not expect to meet “in time” — as a shadow-mapped instance, alongside Google DeepMind’s framework recommending a minimum security level the whole field should apply.
Read against those two definitions, Amodei’s essay is exactly what NIKOLAI’s taxonomy predicts a “pace the frontier” essay would contain: a unilateral Commitment (embedded evaluators, which Anthropic adopted on its own) bundled with an Industry-Wide Recommendation (coordinated pacing, which the essay’s own footnote says needs government mediation or an antitrust waiver to become binding). NIKOLAI does not adjudicate whether the rival CEOs’ public agreement with that recommendation, within the same hour, turned it into something more than a recommendation — that is precisely the question the complaint asks a jury to answer, and it is a legal question, not a taxonomy one. But the distinction NIKOLAI’s N9 track already draws — between a lab saying what it will do and a lab saying what the whole industry should do — is the same distinction this lawsuit’s entire theory of liability depends on collapsing. That NIKOLAI needed a separate element for “industry-wide recommendation” in the first place, built before this complaint existed, is itself a sign the ambiguity was already visible in the safety-commitment landscape it catalogs.
CASRAI’s guide to how ten labs and regulators use “commitment” inconsistently covers the broader pattern of loosely-defined pledges NIKOLAI’s Commitment element was built to standardize; readers assessing litigation exposure from AI safety statements more broadly may also want CASRAI’s general counsel’s guide to frontier-AI litigation exposure.
Frequently asked questions
Have Anthropic, OpenAI, SpaceXAI, or Google responded to the lawsuit?
Unite.AI’s initial report on the filing does not include a statement from any of the four defendants. CASRAI will note any response here once one is publicly available.
Is CASRAI’s NIKOLAI a legal authority on whether this coordination was lawful?
No. NIKOLAI is CASRAI’s own independent, unendorsed reference dictionary for frontier-AI-safety terminology. It defines and tracks concepts like “Commitment” and “Industry-Wide Recommendation” as they appear in labs’ public statements; it does not interpret the Sherman Act, and nothing on a NIKOLAI element page is a finding about this case.
Does the lawsuit challenge Anthropic’s Responsible Scaling Policy itself?
No. The complaint expressly states the plaintiffs do not challenge any defendant’s unilateral decisions about safety, testing, or the pace of its own development — the kind of self-binding pledge NIKOLAI’s Commitment element tracks. The claim targets the alleged coordination among the four companies, not any one company’s own safety framework.
What is “SpaceXAI”?
The complaint names SpaceXAI LLC as a defendant, describing it as the entity through which Elon Musk founded and controls the Grok chatbot business. CASRAI has not independently verified SpaceXAI’s corporate structure beyond what the complaint, as reported by Unite.AI, states.
What relief are the plaintiffs seeking?
Treble damages under the Clayton Act for the proposed class of U.S. paid subscribers to ChatGPT, Claude, Grok or Gemini since September 12, 2026, plus an injunction barring the defendants from horizontal agreements on AI development pace, training-compute limits, coordinated release delays, or exchanges of competitively sensitive information to police such a restraint. The complaint demands a jury trial.
Sources
Unite.AI, Mira Kellan, “Consumers Sue Anthropic, OpenAI, SpaceXAI and Google Over Alleged AI Pact” (September 18, 2026). NIKOLAI element pages verified live at casrai.org/nikolai/element/commitment and casrai.org/nikolai/element/industry-wide-recommendation.
