Jonathan Kanter, the former head of the US Department of Justice Antitrust Division under the Biden administration, says frontier AI companies asking for permission to coordinate on safety do not need an antitrust exemption to do it. Speaking on the Decoder podcast published September 19, 2026, Kanter argued that existing product liability law already forces OpenAI, Anthropic, Google DeepMind, and xAI to ship safe products — and that a formal exemption would function as regulatory capture, not safety policy.
The pitch Kanter is rejecting has been building for months. Anthropic CEO Dario Amodei has publicly used the phrase 'pace the frontier' to describe an industry-wide slowdown, and researchers at multiple labs have quit citing safety concerns. Some of those researchers have put the odds of AI-driven catastrophe above 10 percent. The lab CEOs have paired those warnings with calls for regulation, including carve-outs that would let competitors talk to each other about how fast to move.
Kanter's framing is that the industry has skipped a step. He compared the current moment to inventing cars before painting lane lines, hanging traffic signals, or posting speed limits. His prescription splits the responsibility: companies build safe products, government sets the enforceable rules of the road, and neither job requires the largest players to sit in a room together deciding pace.
“The state of the world right now is like we've invented cars and trucks, but we have no lines on the road, no traffic lights and no stop signs and no speed limits.”— Jonathan Kanter, Former DOJ Antitrust Division chief
Key facts
- 01Jonathan Kanter, former DOJ antitrust chief under the Biden administration, rejects the idea that frontier AI labs need an antitrust exemption to coordinate on safety.
- 02Kanter argues product liability law already covers AI harms — if an AI agent hacks another system, the company that shipped it is on the hook, same as if a human employee did it.
- 03Anthropic CEO Dario Amodei has publicly pushed to 'pace the frontier,' language critics read as a call for industry-wide slowdown coordination.
- 04David Sacks, formerly the Trump administration's AI czar, has publicly aligned with former FTC chair Lina Khan in opposing any AI antitrust exemption.
- 05Kanter's remarks were made on the Decoder podcast published September 19, 2026.
He offered two readings of why the labs want the exemption. The charitable one is genuine fear — engineers who believe their own systems could cause mass harm and want an outside referee. The cynical one is financial: the labs are burning cash at rates that will complicate their IPOs, and a coordinated slowdown would let each of them pull back on spending without losing ground to a rival. Kanter said both readings can be true at once, and neither justifies suspending competition law.
On the sincerity question, he was direct about the safety warnings coming from lab leadership.
“I believe they believe it could destroy humanity. I'm not sure that that doomsday scenario is accurate, but I believe they believe that.”— Jonathan Kanter, Former DOJ Antitrust Division chief
The substance of Kanter's antitrust argument is that liability, not coordination, is the right tool. If an AI agent built by one company breaks into a system owned by another, the company that shipped the agent is legally responsible — the same way an employer is responsible when an employee commits a crime on the job. That principle, he said, does not require Boeing and Airbus to coordinate to keep doors from falling off planes, and it does not require frontier labs to coordinate to keep agents from hacking each other's infrastructure.
He carved out a narrow zone where collaboration is legal today without any exemption: threat-sharing clearinghouses, malicious-bot repositories, and other pooled safety infrastructure that mirrors what banks and airlines already do. Antitrust law does not prohibit that work. What it does prohibit is competitors agreeing to slow product development or cap investment — and that, Kanter said, is what a 'pace the frontier' agreement would look like in practice.
The politics around the exemption question have produced strange bedfellows. David Sacks, formerly the Trump administration's AI czar and an avowed libertarian, has publicly amplified former FTC chair Lina Khan's position that no exemption is needed — a rare cross-ideological alignment. Meanwhile the Trump DOJ has kept the antitrust case against Apple alive, which surprised observers who expected a full reversal of Biden-era enforcement. The Google case Kanter brought during his tenure resulted in a government win, as did the Ticketmaster action.
“If two companies say, "We're competing too hard and we need to slow down," yes, that could implicate the antitrust laws, but that's not what we should be doing here.”— Jonathan Kanter, Former DOJ Antitrust Division chief
The counterweight to Kanter's view comes from inside the labs themselves. Amodei and others argue that unilateral safety investment is a competitive disadvantage — the lab that spends most on alignment ships slower and loses market share to a rival that spends less. In that framing, the prisoner's dilemma is real, and product liability arrives only after the harm, not before. Kanter's response is that this is exactly what regulatory line-drawing is for, and it does not require the regulated parties to write the lines themselves.
The practical stakes for the AI market are significant. If Kanter's reading holds inside the current DOJ, frontier labs will not get a legal shield for pace-setting talks, and any coordination they attempt will carry real antitrust exposure. That pushes safety work back into two channels: unilateral company investment, and government-set rules with teeth. For investors watching pre-IPO burn rates at OpenAI and Anthropic, it also removes a convenient narrative — the industry cannot argue that a coordinated slowdown is a safety necessity when the former top antitrust enforcer in the country is on record saying it isn't.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
