The frontier AI labs want permission to coordinate. Jonathan Kanter, who ran the DOJ Antitrust Division under Biden, says they shouldn’t get it. In a new Decoder interview with The Verge AI, Kanter walks through why calls for an antitrust exemption on AI safety don’t hold up, whether you read the labs’ motives generously or cynically.
This matters because the exemption ask has become the loudest policy request in AI right now. OpenAI, Anthropic, Google DeepMind and Elon Musk’s xAI have all signaled they want to slow down, with Anthropic’s Dario Amodei calling it a plan to “pace the frontier.” The catch: coordinating on pace among competitors is textbook cartel behavior unless the government blesses it.
What Kanter actually said
Kanter’s framing is simple. “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.” He thinks the industry needs rules. He doesn’t think the industry should write those rules for itself under legal cover.
He lays out two readings of the exemption push, according to The Verge AI:
- The generous version: The labs genuinely fear the pace of development and want government guardrails. Kanter says he believes they believe the doomsday risk is real, even if he’s not convinced himself.
- The cynical version: The labs are burning cash at an unsustainable rate. Pulling back alone would tank their valuations ahead of IPOs. An exemption lets everyone slow spending together without anyone looking weak.
His conclusion applies to both: neither justifies an antitrust exemption. If the government wants to regulate AI, it can regulate AI. Competitors don’t need a legal shield to lobby for that.
Why the timing is unusual
The backdrop here is a Trump administration that has mostly declined to regulate frontier models, plus a wave of safety researchers leaving Anthropic and DeepMind publicly. Some researchers now put the odds of catastrophic outcomes above 10 percent. The labs are responding by asking for coordination rights, and critics see regulatory capture, a cartel in formation, or a graceful exit from investor pressure.
What stands out is the coalition forming against the exemption. David Sacks, a libertarian and former Trump AI czar, has been sharing Lina Khan’s position that no exemption is needed. When Khan and Sacks agree on competition policy, something unusual is happening. It suggests the exemption idea has few natural allies outside the labs themselves.
Kanter’s track record adds weight. He won the DOJ cases against Google and Ticketmaster, and his Apple case is still alive under the Trump DOJ. He’s not a theorist on this. He’s the person who spent four years proving that tech coordination and market dominance are prosecutable.
The China angle
The interview also touches on a quieter concern: whether a coordinated slowdown among US labs would foreclose competition from cheaper open-weight Chinese models. If the big labs agree to pace the frontier, and the government enforces that pace, open-weight rivals could end up as collateral damage. That’s a competition problem dressed up as a safety measure, and it’s exactly the kind of outcome antitrust law exists to prevent.
What this means for you
If you build on frontier models or run a business that depends on them, here’s the practical read:
- Don’t expect a coordinated slowdown. The legal path for one doesn’t exist yet, and the politics are against it. Model releases will keep coming at current pace.
- Watch for regulation, not exemptions. Kanter’s position points toward the government setting rules directly. That’s slower but more likely than a lab-led pact.
- Hedge on open weights. If US labs do find a way to coordinate, open-weight models from China and elsewhere become the alternative. Keep your stack portable.
- Read IPO signals carefully. The cynical interpretation deserves attention. Slowdown talk that coincides with pre-IPO positioning tells you something about unit economics.
Looking ahead
Over the next one to two years, the real fight won’t be over whether the labs can coordinate. It’ll be over whether Congress or the executive branch writes rules at all. Kanter’s view is that the responsibility runs both ways: companies have obligations, and so does government. Right now neither side is meeting them.
This is part one of a two-part Decoder series. The full conversation goes deeper on how safety regulation would interact with competition law and the US-China dynamic. Worth the full listen at The Verge AI.