Breaking

U.S. Antitrust Law May Block AI Labs From Coordinating a Superintelligence Slowdown

September 14, 2026 at 20:00 EDT

Policy & Regulation
Foundation Models

U.S. antitrust law could treat an agreement among frontier AI companies to produce less superintelligence as an illegal output restriction, even if the stated purpose is to reduce the risk of human extinction. That tension is now at the center of an industry campaign to slow the most advanced systems, after Anthropic, OpenAI, and other labs called for coordinated pacing of capabilities they say are accelerating beyond human control.

September 2026 · Frontier AI

The slowdown that could look like a cartel

Frontier labs want to coordinate less superintelligence to cut extinction risk. U.S. competition law may still treat that pact as an illegal output restriction.

$30T

Anthropic TAM cited for IPO investors

1–2 yr

Extra time Amodei says alignment needs

6–12 mo

Window he flags for a catastrophic swarm

3 steps

Evaluators, then democracies, then a global deal

How the pacing plan is supposed to work

Step 1

Embedded evaluators

Third-party staff inside each lab. Anthropic is starting unilaterally.

→

Step 2

Democratic coordination

Shared safety standards and pacing limits. Needs an antitrust waiver.

→

Step 3

Global coordination

Graded deals with China. A full pause is called unlikely soon.

The U.S. lead a pause would rely on has already shrunk

Reported lead of U.S. models over foreign models, before and after distillation attacks.

12–18 mo.

Prior U.S. lead

4–6 mo.

After distillation

Takeaway: Distillation has cut the reported U.S. lead to about one-third of its former length.

Safety case

A verified pause could buy one to two years for alignment. Embedded evaluators can publish findings. The first step needs no waiver.

Cartel critique

An output limit among incumbents could freeze today's hierarchy. Smaller and open-weight labs may be boxed out. Washington and Beijing have rejected a pause.

The legal bind

Sherman Act Section 1 can treat a safety pact as an output conspiracy

Unilateral pause: low antitrust risk, little effect on the race.
Formal pause among rivals: typically illegal per se as an output restriction.
A statutory mandate or narrow waiver is the path that would largely clear the antitrust problem.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.