Do AI companies need special permission to work together safely?
Major AI companies are asking for antitrust exemptions—special permission to coordinate on safety. A former US antitrust chief explains why he thinks that's the wrong approach.
The Ask: Special Permission to Team Up
OpenAI, Anthropic, Google DeepMind, and other leading AI companies have asked the U.S. government for something unusual: an antitrust exemption. This is special permission that would allow these normally competing companies to work together on safety issues without breaking competition laws. Their argument is straightforward—AI development is moving so fast, and the stakes are so high, that they need to coordinate to prevent things from going wrong.
But Jonathan Kanter, who until recently was the head of the U.S. Department of Justice's antitrust division (the government office that prevents companies from unfairly controlling markets), thinks this is a bad idea. "These companies do not need to coordinate in order to deliver safe and secure products," he told The Verge in an interview. His point: just like Boeing doesn't need permission to coordinate with Airbus to fix plane door problems, AI companies don't need an exemption to build safer products.
What He Thinks Should Happen Instead
Kanter's prescription is simpler: hold companies responsible for what their AI systems do. If an AI agent breaks into someone else's computer, that's the company's fault—just like if an employee did it. Companies should be liable for harmful outcomes, similar to how traditional product liability laws work (the framework that makes a car manufacturer responsible if their brakes fail). Congress should also clarify what society values—privacy, competition, accurate information—and require companies to build those values into their products from day one.
The deeper issue, according to Kanter, isn't that companies need exemptions. It's that the U.S. government hasn't set clear "rules of the road" for AI, the way it did for cars, planes, and medicine. Congress has been unable to act effectively due to political gridlock. Meanwhile, AI companies are delivering powerful tools without adequate safety built in—repeating the same mistakes we saw with social media.
Original source: The Verge AI