Sacks to OpenAI and Anthropic: Slow AI if You Must, but Drop the Regulation Demands

David Sacks, the Trump administration’s AI point man, publicly answered calls from Anthropic’s Dario Amodei and OpenAI’s Sam Altman to “pace the frontier” of AI development. His retort: go ahead and slow down — but stop attaching conditions. Sacks argues the two companies already hold a duopoly over frontier intelligence by every metric that matters (market share, revenue growth, capability) and have themselves claimed their lead is widening through recursive self-improvement. If their internal, unreleased models genuinely alarm them, he says they are free to act responsibly on their own without anyone else’s sign-off.

The sharper point is his rejection of the regulatory scaffolding the labs want in exchange. Sacks tells them to stop implying they need antitrust immunity to coordinate, a licensing regime that overrides ordinary product liability, or third-party evaluators to police competitors who haven’t even reached the frontier. He singles out METR, arguing it isn’t truly independent given its ties to Anthropic’s investors and staff. He also disputes that the motive is pure altruism: after the “Hugging Face episode,” trading raw capability for reliability is simply sound business, since a model implicated in a serious cyberattack would carry heavy liability and markets punish unpredictable systems. Call it alignment, he says, but it’s also just meeting customer demand.

Sacks frames the whole exchange as a test of intent. Pacing development, he concedes, could buy room for a smarter policy debate than “shut it all down,” and he notes China is unlikely to sign any global pact. But if the labs treat their preferred regulatory framework as the price of restraint, he warns, it will read as blackmail of the public and a bid for regulatory capture — or, given the timing, an election-season stunt. The message: you set the frontier, so you can pace it, but do it without capturing the rulebook.