The debate over whether AI should slow down has moved from tweets and keynote stages into a courtroom. A new civil lawsuit alleges Anthropic, OpenAI, SpaceXAI and Google made an illegal agreement to coordinate the pace of their AI development — four AI giants now face antitrust claims over their calls to "pace the frontier."
The suit was brought by four paid subscribers — users of ChatGPT, Claude, Grok and Gemini respectively — and filed Friday, September 19, in the US District Court for the Northern District of California. One of the plaintiffs' attorneys, Nick Rowley, said the case was filed "to prevent the world's most powerful for-profit tech companies from conspiring, without consumers' or regulators' knowledge, to control the pace of this epochal technology."
The complaint's core evidence is last week's high-profile "slowdown coalition." It argues the coordination largely took place on September 12 — the day Anthropic CEO Dario Amodei published his essay urging industrywide cooperation to slow capability gains in favor of safety, with OpenAI's Sam Altman, SpaceXAI's Elon Musk and Google DeepMind's Demis Hassabis endorsing it within hours. The filing contends these acts constituted an illegal business agreement between competitors: they "specified that their development pace should be slower than it would be under full competition," which is anticompetitive and reduces the value consumers get from paid AI subscriptions. It also notes the coordination had been brewing for months, citing a July statement by AI employees calling for a global slowdown.
The lawsuit pushes the AI safety debate into a new dimension: when "safety coordination" becomes a charge of "collusive deceleration," Amodei's own preemptive concern about antitrust comes full circle — his essay acknowledged that industry coordination would need government mediation or antitrust waivers. Ironically, OpenAI had confirmed the same day that it is building a FINRA-style standards body with Anthropic and Google DeepMind, which Cohere's CEO has publicly called "a cartel." Now "cartel" has moved from criticism to a formal allegation in court filings.
The four companies have not yet responded. Legal experts note the difficulty lies in distinguishing legitimate industry safety coordination from an illegal horizontal agreement — and the nearly synchronized public endorsements of September 12 give plaintiffs a ready-made evidence chain. Whatever the outcome, the suit sends a warning to the AI industry: any industry coordination under the banner of "safety" may also trip antitrust red lines. The second half of the slowdown debate will be decided by judges, not tweets.