What Happened?
If leading AI labs jointly slow powerful AI for safety, they could face lawsuits for breaking competition law. That bind leaves rival developers of frontier models with no settled path to jointly pause work without inviting unlawful-restraint claims, a new Congressional Research Service report finds.
On Sept. 12, Anthropic chief executive Dario Amodei urged peers to pace advanced systems and proposed shared standards and limits that could need government backing including waivers. On Sept. 18, customers of several leading companies filed a proposed class action alleging an agreement to slow innovation.
Section 1 of the Sherman Antitrust Act (Sherman Act), the law against contracts that restrain trade, bars every such contract but courts read it to ban only unreasonable harms. Price fixing, market division and output limits are condemned without inquiry into effects, while most other restraints face fact-specific inquiry into harm such as higher prices, reduced output or diminished innovation.
Why Does it Matter to Me?
Competing developers of frontier artificial intelligence models have no settled path to jointly slow work for safety without inviting unlawful-restraint claims. A coordinated halt above capability thresholds carries the greatest exposure as a horizontal curb on quality and research competition.
Just today, President Trump met with the CEOs of top AI companies in Washington DC to applaud their progress. There was no talk of regulation or slowing these companies down.
Joint testing of prerelease models may pose heightened risk. Sharing cybersecurity incidents alone presents likely low risk, sharing alignment methods presents low to moderate risk, and using outside evaluators could reduce swapping sensitive information.
Private standard setting is more defensible where standards rest on objective expertise through unbiased procedures, though joint enforcement faces stricter review.
Both Sides, Now
Other lab leaders endorsed pacing, and House Minority Leader Hakeem Jeffries sought congressional action. Amodei tied the call to curbing loss of control and misuse for cyberattacks and bioterrorism.
Critics said current law already permits much safety work and waivers could harm competition. Separate commentary cited rivalry with the People's Republic of China and framed a slowdown as pretext to lower capital outlays.
Congress controls whether to create a legal shield, and in the 119th Congress the Collaboration on Adversarial Threats and Security Risks Act (CATS Act) would excuse good-faith exchange on covered security risks and coordinated delay upon written notice to the Antitrust Division head. That bill is structured as an affirmative defense proved by preponderance, while narrower choices include an information-only excuse or renewal of cyber sharing cover and H.R. 9925 would build Commerce oversight.
What Happens Next?
Protection under the Cybersecurity Information Sharing Act of 2015 (CISA 2015), the law that shields cyber threat sharing, ends Dec. 11 absent extension. Research and development guidance from 2000 was withdrawn in December 2024 as no longer reliable guidance, though case law still supports review under the rule of reason.
A July 2025 statement distinguished arrangements leaving firms free to sell nonconforming products from those forbidding them, and a 2025 inquiry into truck makers honoring California emissions rules closed in August 2025 after pledges to act independently. Lawmakers confront competing designs for broad exemptions versus narrow exemptions versus upfront regulation, and a pending court test of whether safety justifications survive precedent.
---
AllCiv makes it easy to stay politically informed and involved.
---
Spot something wrong? Report an issue with this article