Four consumers filed a class action in the Northern District of California on September 18, 2026, naming Anthropic, OpenAI, SpaceXAI, and Google. Buist v. Anthropic PBC alleges a horizontal agreement to slow the rate at which the four improve competing products, contrary to section 1 of the Sherman Act. The lawsuit seeks treble damages, an injunction, and a jury trial.
How we got here
The immediate trigger was a September 12 essay by Anthropic CEO Dario Amodei, who urged leading AI firms to cooperate on slowing capability advances and devote more time to safety. He suggested the U.S. government could grant a narrow antitrust waiver for such discussions.
On the same day, OpenAI CEO Sam Altman, SpaceXAI CEO Elon Musk, and Google DeepMind co-founder Demis Hassabis publicly responded in agreement — though with important nuances. Amodei wrote that coordinating on pace would need an antitrust waiver. Altman said OpenAI would not wait for one. OpenAI policy chief Chris Lehane later confirmed the company had already been working with Anthropic and Google DeepMind for several weeks.
Those public statements are now evidence in a courtroom, not just debate on social media.
What the plaintiffs allege
The pleaded term is a slower rate of capability advancement than each company would choose alone. The mechanisms alleged include limits on training compute and training runs, restrictions on using AI to improve AI, and capability checkpoints described as verifiable. That language mirrors safety programs Anthropic has discussed publicly — which the complaint uses as circumstantial evidence of coordination.
The antitrust exposure was discussed openly before anyone filed. A statement from employees across the labs in July said each faced intense competitive pressure not to unilaterally slow, which the complaint now uses as evidence of both motive and action against self-interest.
The political context
The Trump administration has been vocal about wanting American AI labs to outpace Chinese competition. On September 19, Trump announced plans for an "AI czar" and an "AI commission," though he provided no details. In a social media post, he said the U.S. would not "hinder or stifle" the growth of the AI industry.
Republicans have largely mirrored that position, while several Democratic leaders have called for sweeping AI regulation. The lawsuit sits uncomfortably at the intersection of competition law, national security rhetoric, and genuine safety concerns.
Why this case is hard to predict
Antitrust claims require evidence of a concrete agreement, not just parallel public statements. Legal analysts note that executives agreeing on the importance of safety is not the same as agreeing to fix prices or reduce output in a way that harms consumers. Plaintiffs will need to show that coordinated slowdown efforts reduced the value subscribers receive from paid AI products.
Defendants will argue that safety coordination protects users and that no illegal agreement exists. They may also point to continued model releases — OpenAI's GPT-6 Astra and Agents API launched this month — as evidence that competition remains intense.
What developers and users should watch
The case will not resolve quickly, but its discovery process could force unprecedented disclosure about how frontier labs evaluate capability thresholds, internal red-teaming, and cross-company safety meetings. For developers building on these APIs, the near-term impact is minimal. For anyone shaping AI policy, the lawsuit makes explicit a tension the industry has managed privately for years: safety coordination looks like collusion when viewed through antitrust lenses.
Watch whether the alleged working group continues meeting. The complaint alleges it was still meeting in mid-September. Continuing now would be read as continuing the conduct — with legal and reputational consequences either way.
Comments
Loading comments…