The sharpest fault line in American AI policy cracked open further on Monday, August 17, 2026, when White House AI and Crypto Czar David Sacks published a nine-point rebuttal to Anthropic CEO Dario Amodei, accusing him of advancing a regulatory framework that would hand a small circle of frontier labs — and a federal bureaucracy — gatekeeping power over which AI capabilities reach the public.

The Straw Man Sacks Won't Accept

Amodei had argued that critics of AI regulation live in a 'bubble' where all oversight is automatically suspect. Sacks dismissed that framing as a straw man. 'Of course treating all regulation as capture would be overly simplified — but almost no one holds that view,' he wrote, noting that he has himself called for strong antitrust enforcement against Big Tech and would apply those rules to Anthropic too if the company drifted toward monopoly status.

The distinction matters because it moves the debate off a cartoon and onto the specific proposal Amodei has championed: a new federal agency that would review and approve frontier AI models before release. Sacks has heard it pitched as an 'FDA for AI,' an 'FAA for AI,' and most recently a 'FINRA for AI.' He calls it a 'DMV for AI' — a queue-generating bureaucracy whose rules would only grow more labyrinthine over time.

Stigler's Ghost and the Revolving Door

Sacks reached for Nobel laureate economist George Stigler to give the charge of regulatory capture its precise meaning. Stigler defined it as regulation acquired by an industry and designed primarily for that industry's benefit — a direct challenge to the idea that agencies arise from a benevolent state correcting market failures. The mechanism, Sacks argued, is well understood: concentrated industry stakes, sustained lobbying, and the revolving door between companies and the agencies that oversee them.

He applied that framework directly to Anthropic, which he said has hired multiple senior Biden administration AI-policy officials and built a substantial government-affairs operation alongside a network of aligned organisations pushing its preferred frameworks at state and federal levels. The company, he argued, is structurally positioned to shape any approval process it nominally submits to — while smaller competitors wait in line.

The China Gap and the Open-Model Question

The geopolitical stakes sharpened the argument. China, Sacks noted, will not adopt equivalent pre-release approval requirements. A U.S. licensing regime would therefore handicap American developers relative to Chinese competitors who face no such queue — a risk he said is compounded by Anthropic's posture toward open-source models. Amodei has called open models dangerous in Senate testimony, criticised their lack of central monitoring, and linked them to intellectual-property theft. Sacks argued that insisting identical rules apply to both open and closed models could achieve a de facto ban without ever calling it one, leaving the United States 'an island of costly closed models while the rest of the world races ahead.'

Anthropic is on track to become one of the most valuable companies in history, Sacks wrote, with the resources to navigate any approval process and influence the rules while rivals wait. That asymmetry, he said, is not incidental — it is the structural logic of the proposal.

Fear Narratives and the 'Vision of the Anointed'

The final section of Sacks's post turned to public messaging. He disputed Amodei's claim that negative public sentiment toward AI stems from a long-standing erosion of trust in institutions, arguing instead that Anthropic's own communications — including a forecast, which Sacks cited but which remains unverified, that AI would eliminate 50 percent of entry-level knowledge jobs within five years — have done more than anything to shape public fear.

He invoked economist Thomas Sowell's concept of 'The Vision of the Anointed' to describe what he sees as the underlying mindset: elite intellectuals convinced that only they possess the judgment to govern a powerful technology. And he closed with investor Gavin Baker's crisp summary of the divide: 'Dario believes frontier AI is too powerful to distribute; we believe it is too powerful to centralize.' The weight of human history, Sacks concluded, gives reason to fear the outcome of concentrating that authority in a marriage of corporate and state power.

That sentence is not a rhetorical flourish. It is a direct policy challenge — and Washington will have to answer it.