A lawsuit filed Friday in the U.S. District Court for the Northern District of California alleges that Google, OpenAI, SpaceXAI, and Anthropic entered into an illegal agreement to slow the development of artificial intelligence. The plaintiffs argue that this coordination violates federal antitrust laws and reduces the value consumers receive from paid AI subscriptions.
The legal action centers on events from Sept. 12, when Anthropic CEO Dario Amodei published an essay calling for industry-wide cooperation to decelerate AI advancements to improve safety. According to the filing, the leaders of OpenAI, SpaceXAI, and Google DeepMind confirmed their agreement to this proposal on the same day. Amodei has stated that rapid progress necessitates a more cautious approach to prevent AI agents from potentially disrupting the internet.
Attorneys representing four named plaintiffs—users who pay for ChatGPT, Claude, Grok, or Gemini—are seeking class-action status for all paid subscribers of these services. The lawsuit claims that competitors are not legally permitted to decide among themselves that market competition is too dangerous. Lead attorney Nick Rowley stated that allowing AI safety and protocol to be controlled by private agreements could result in AI spinning out of human control.
In his original proposal, Amodei acknowledged that such coordination might face antitrust challenges. He suggested that the U.S. government could mediate the discussions or issue a "narrow waiver" to allow specific safety-related conversations between rival labs. Amodei told CBS News that while he believes the probability of negative outcomes is low if developed correctly, the risks are high if the industry proceeds in the "wrong way." Representatives for the four companies did not provide an immediate response to the allegations on Saturday.
The scale of the impact involves four technology companies and a proposed nationwide class of paid subscribers. A legal ruling against the companies could set a precedent regarding whether "safety" concerns can serve as a legal defense for what would otherwise be considered anti-competitive behavior. For the average user, this case may determine whether the pace of new feature releases is dictated by market competition or by private agreements between the providers.
The outcome of this case will also influence how the federal government interacts with the AI industry. If the court rejects the companies' safety arguments, it may force them to compete more aggressively on speed. Conversely, if the companies prevail, it could establish a framework where tech firms are permitted to coordinate on development timelines if they claim a safety justification. The case is currently in the U.S. District Court for the Northern District of California; no trial date or deadlines for response have been reported.