OpenAI asked Congress whether an AI industry slowdown would be legal under antitrust rules
WIRED ● Covered by 85 sources
OpenAI asked Congress if AI labs can legally slow down together. The twist: safety talks may run into antitrust law before they run into the tech.
Based on reporting by WIRED — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
OpenAI has been quietly pressing lawmakers for a simple answer to a messy question: if AI companies agree to slow frontier model work for safety, does that break antitrust law? People close to the company say the concern is real enough that it could scare off the big players OpenAI wants at the table.
That debate got louder after OpenAI chief scientist Jakub Pachocki published a blog post last weekend arguing that the AI research world should be “coordinating to slow down future development.” He says voluntary slowdowns may become normal until the field sets shared safety bars. The logic is straightforward. If systems can improve themselves, the pace of development matters as much as the raw capability race.
But the law may not like the shortcut. Nicholas Felstead, a former AI policy fellow at the Center for Law & AI Risk, wrote in March that a coordinated pause could look like companies restricting output, which might violate the Sherman Antitrust Act. He also argued that even collaborations with a good safety case can be chilled by uncertainty. That’s often how these fights go: the legal ambiguity does half the work before any court does.
Congress is at least poking at the problem. In July, a bipartisan, bicameral group introduced the Collaboration on Adversarial Threats and Security Risks Act, which would let AI labs coordinate on security and safety without tripping antitrust rules. The House version has been sent to the Judiciary Committee, but it hasn’t moved yet. Caleb Knapp of the AI Policy Network, which backs the bill, says lawmakers have a growing appetite to act, though anything real may wait until after the midterm elections.
Still, antitrust may not be the whole story. AI companies are fighting over a huge market, and some executives also see the race with China through a national security lens. On top of that, labs don’t agree on what “safe” even means. OpenAI cofounder John Schulman, now at Thinking Machines, said the problem is less about antitrust than OpenAI and Anthropic not agreeing on a pacing proposal in the first place. Meanwhile, a string of security incidents and a public warning from former Anthropic and OpenAI researcher Jacob Coxon have made lawmakers more eager to regulate the field anyway.
My take — AI-written commentary, not fact-checked reporting
The antitrust excuse is doing a lot of heavy lifting here. If AI leaders want to coordinate on safety, they should say exactly what they mean and stop hiding behind legal fog and noble-sounding panic. Congress is being asked to bless an industry truce, which is rich coming from a sector that can barely agree on the definition of safe.
Read more about this at: WIRED
