Rapid News Brief: Four paid AI subscribers filed a proposed federal class action on Sept. 18 against Anthropic, OpenAI, SpaceXAI and Google, alleging that the companies illegally coordinated to slow the pace at which their competing AI products improve.
The complaint, filed in the U.S. District Court for the Northern District of California, names Claude, ChatGPT, Grok and Gemini subscribers as the proposed class. It alleges that an agreement to pace frontier-model development reduced the quality and rate of improvement customers expected from paid subscriptions.
A safety proposal becomes an antitrust claim
The case, Buist v. Anthropic, PBC, is docketed as No. 3:26-cv-10693. The plaintiffs say a Sept. 12 essay by Anthropic CEO Dario Amodei proposed industrywide coordination and that public responses from leaders at OpenAI, SpaceXAI and Google DeepMind showed agreement.
Amodei’s essay argued that frontier-model capabilities should advance more slowly so safety work could keep pace. It also acknowledged that some forms of company-to-company coordination would be legally difficult and could require government support. The complaint treats that coordination as an alleged restraint on product improvement under Section 1 of the Sherman Act.
The claims remain unproven
The filing is an allegation, not a judicial finding. The proposed class has not been certified, and the court has not ruled that public endorsements of a slower development pace created an unlawful agreement. The Associated Press reported that none of the four companies immediately responded to requests for comment Saturday.
The plaintiffs seek class certification, treble damages under the Clayton Act and an injunction against agreements covering model-development pace, release timing, training limits and capability checkpoints. Bloomberg Law separately confirmed the filing and case number.
Coordination now carries a new business risk
For frontier labs, the immediate consequence is not a court-ordered change but a higher documentation and governance burden around joint safety work. Companies may need to distinguish unilateral controls from shared commitments and route industry discussions through formal standards processes or government-backed frameworks.
Enterprise buyers should watch whether vendors change how they describe safety roadmaps, release timing and cross-company standards. The suit could also push regulators to clarify when competitors may cooperate on AI safety without constraining competition. Until the defendants answer and the court tests the complaint, the alleged agreement and claimed consumer harm remain disputed.