OpenAI, Anthropic, xAI, and Google Sued Over Alleged AI Collusion
Four consumers have filed a class-action antitrust lawsuit against OpenAI, Anthropic, xAI, and Google, alleging the companies conspired to intentionally slow down AI model improvements.
A proposed class-action lawsuit, Buist v. Anthropic PBC, has been filed in federal court against OpenAI, Anthropic, xAI, and Google. The plaintiffs, Charles Buist, Nick Spetsas, Christine Bullock, and Cheyenne Hunt, allege that these companies violated Section 1 of the Sherman Act by coordinating to deliberately throttle the development of their AI models.
The lawsuit claims that these four companies control approximately 80% of the paid consumer subscription market for frontier AI models, including ChatGPT, Claude, Grok, and Gemini. The plaintiffs argue that paying customers were promised cutting-edge tools that would continue to improve, but that the companies instead reached a secret agreement to slow down progress.
The complaint points to an essay published on September 12 by Anthropic CEO Dario Amodei, which advocated for a coordinated deceleration of AI capability improvements. The plaintiffs note that OpenAI CEO Sam Altman and Elon Musk of xAI publicly endorsed this sentiment shortly after the essay was released.
Additionally, the lawsuit highlights a meeting in July where representatives from Anthropic, OpenAI, and Google reportedly participated in a working group to establish an industry standards body. While the companies may argue these discussions were focused on safety benchmarks, the plaintiffs characterize the meeting as the foundation for an illegal agreement to limit output.
The plaintiffs are seeking class certification for all US paid subscribers to the defendants' AI services. They are requesting injunctive relief to stop the alleged coordinated slowdown and a declaratory judgment that antitrust laws were violated. The current complaint does not specify monetary damages.
Legal experts note that antitrust law does not provide exceptions for safety-related intentions if competitors agree to limit product quality or development speed. However, courts have historically been skeptical of antitrust claims that rely on public statements and inferred agreements rather than direct evidence such as emails or contracts. The outcome may depend on whether discovery reveals more than general discussions regarding safety protocols.
