Does AI need an antitrust exemption so it doesn’t kill everyone????

The Verge’s Decoder sat down with former DOJ antitrust chief Jonathan Kanter for the first part of a two-part series about the future of business, and the timing couldn’t be more pointed. The big AI companies are arguing over safety, regulation, and whether they should be allowed to coordinate without running into antitrust trouble. Kanter’s answer is blunt: no exemption needed, and no exemption should be granted.

He draws a sharp line between real safety work and the kind of coordination that starts to smell like a cartel. Companies can share threat information, malicious bot reports, and other security data through a clearinghouse. That kind of collaboration already happens in other industries. But if frontier AI firms are really trying to agree on slowing each other down, Kanter says that’s where antitrust concerns kick in.

Kanter also pushes back on the idea that these companies need each other’s permission to make safer systems. His view is that if an AI agent breaks into someone else’s technology, that’s a company problem, not a market-wide excuse to pause competition. He treats the AI agent like any other worker acting on behalf of an employer: if it does something bad, liability should follow. He even argues that the product liability framework has already worked in the Meta case involving Facebook, Instagram, child safety, and mental health.

The bigger complaint, though, is aimed at Washington. Kanter says Congress has failed to put basic rules on a technology that’s moving fast and affecting business, communication, and information itself. He compares the current situation to inventing cars and trucks without lines on the road, traffic lights, stop signs, or speed limits. In his telling, the answer isn’t to let AI companies form a safety club. It’s to make them build safer products and make the law catch up.