Four consumers filed a class action in the Northern District of California on 18 September, 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.
Bloomberg Law reported the filing. It seeks treble damages and an injunction, and demands a jury.
The waiver sequence is the spine of the case
Amodei’s essay said government mediation would help and that a narrow waiver would be needed for certain safety conversations. The complaint’s answer is flat: no such waiver exists, Congress has enacted no exemption, and no agency has compelled the conduct.
Two days later Altman said OpenAI did not intend to wait for an antitrust exemption or legislation. The following day OpenAI’s policy chief Chris Lehane confirmed the company had already been working with Anthropic and Google DeepMind for several weeks.
The complaint files that sequence under consciousness of antitrust risk. Asking whether the conduct was lawful is pleaded as proof the companies understood it as coordination, and proceeding anyway as proof of intent.
No leaked emails, and that is the point
Paragraph two puts it plainly: the agreement was “proposed in public, accepted in public, and confirmed in public”. There are no internal documents, no confidential witnesses and no meeting minutes in the filing.
Every factual source is a named executive’s public statement or a press report. TNW covered the essay that the complaint treats as the offer, and the response that it treats as acceptance.
That is unusual for a cartel case and it cuts both ways. It makes the evidence easy to verify, and it means the conduct alleged is conduct nobody tried to hide.
What is said to have been agreed
The pleaded term is a slower rate of capability advancement than each company would choose alone. The mechanisms alleged are limits on training compute and training runs, restrictions on using AI to improve AI, and capability checkpoints.
The policing mechanism alleged is the embedded evaluator proposal itself, described as verifiable pacing. That is the same programme Anthropic has begun implementing with Accenture.
What the complaint does not say
It does not allege the safety concerns are invented. Its pretext argument is structural rather than personal, holding that every stated safety objective can be pursued unilaterally, through independent evaluators, or through regulation.
The plaintiffs say they take AI risk seriously, and paragraph eleven describes independent safety investment and welcoming government oversight as the right path. They also disclaim any challenge to unilateral slowing, independent evaluators, lawful safety research, compliance with government requirements, or petitioning Congress, including petitioning for an exemption.
The legal move is narrower than it first appears. Competition being dangerous, the filing argues, is not a defence a court may weigh, because that judgment belongs to Congress.
The weaknesses are visible on the face of it
The complaint concedes that the full effect on released products has not yet manifested. No slowed release, cancelled model or altered training run is identified anywhere in twenty-nine pages.
The claimed 80% share of US paid subscriptions is pleaded on information and belief and as a floor rather than a measurement. The plaintiffs also plead the restraint as naked, which would make market definition unnecessary, and offer the market only in the alternative.
The four acceptances are not equivalent either. Musk’s was four words on social media, and Google’s was an endorsement of the direction with implementation details left open.
Why this was foreseeable
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.
That line is doing a lot of work. An admission that unilateral restraint is commercially irrational is, in antitrust terms, an explanation of why an agreement would be needed.
The commercial context is not incidental. TNW has written about the labs warning of AI risk while racing towards public listings.
Europe would ask the same question
Article 101 of the EU treaty prohibits agreements between undertakings that restrict competition, and contains no safety exemption a company can grant itself. A coordinated pacing arrangement among frontier developers would face the same analysis in Brussels.
There is no European case, and none has been signalled. The point is that the waiver problem Amodei identified is not a quirk of American law.
Geopolitics is already pulling the other way. Beijing called the slowdown plan a Cold War script, which leaves coordination contested at both ends.
What to watch
Watch whether the working group meets again. The complaint alleges it was still meeting in mid-September, and continuing now would be read as continuing the conduct.
Watch the answers. These are allegations filed the day before this was written, no defendant has responded, and the companies have strong arguments that public advocacy and unilateral safety work are lawful.