The U.S. is consolidating AI governance under a single executive while China runs laps with coordinated state planning.
The Summary
- Trump is forming an "AI Force" and appointing a czar to centralize oversight of the AI industry, rejecting independent federal regulation
- The president called himself "the only guardrail" AI needs, publicly attacking Anthropic CEO Dario Amodei for advocating safety controls
- The Defense Production Act, which could grant Cold War-era powers over AI, is set to expire as this power grab unfolds
- Xi's China is outpacing the U.S. on coordinated AI strategy, according to FT analysis, while Trump's anti-regulatory stance ignores the catching-up threat
The Signal
The U.S. is about to run one of the messiest experiments in tech governance: consolidating all AI oversight under presidential control instead of building independent regulatory capacity. Trump's announcement of an AI Force with a single czar reporting directly to him represents the most significant restructuring of AI policy architecture since the sector emerged. This isn't about streamlining bureaucracy. It's about controlling who gets compute, who gets funded, and who gets regulated.
On Truth Social, Trump dismissed calls for AI safety regulation, writing that a "strong, smart president" is all the guardrails the technology needs. He specifically attacked Amodei, calling out the Anthropic CEO by name for advocating development slowdowns and tighter controls. The message to the industry is clear: align with the administration or get labeled an obstacle to American competitiveness.
"The consolidation of AI oversight under presidential control could centralize power, impacting regulatory independence and global competitiveness."
But here's the timing problem. The Defense Production Act, a Korean War-era law that could give the government extraordinary control over AI resources, is expiring just as this centralization accelerates. The DPA would let the president compel AI companies to prioritize government contracts, restrict exports, and redirect compute infrastructure for national security purposes. Losing that tool while concentrating power in a political appointee creates a strange gap: maximum political control, minimum legal teeth.
The real risk isn't over-regulation. It's under-regulation at exactly the wrong moment. Industry leaders including those at Anthropic, OpenAI, and Google DeepMind have been signaling that current development pace is outrunning safety testing. They're not asking for bureaucracy. They're asking for structured evaluation protocols before deployment, particularly for models that will control infrastructure, make financial decisions, or operate with minimal human oversight.
Meanwhile, China is playing a different game entirely. Xi's government has:
- Integrated AI development into national industrial policy
- Coordinated compute infrastructure buildout across provinces
- Created regulatory frameworks that move fast but with clear lanes
- Tied AI progress directly to geopolitical strategy
The FT analysis is blunt: Trump's anti-regulatory posture isn't accelerating American leadership. It's creating confusion about what's allowed, what's coming, and who's in charge beyond the next election cycle.
The Implication
Watch who gets named AI czar and what their actual authority turns out to be. If it's a political loyalist without technical depth or independent statutory power, this becomes performance art. If it's someone with real relationships in the agent economy and the authority to coordinate across Defense, Commerce, and Treasury, it could work. But centralizing control without building institutional knowledge is how you get blindsided by the country that spent the last five years building coordination mechanisms you mocked as bureaucratic.
The companies building Web4 infrastructure need clarity on compute access, export controls, and what "acceptable use" means when your agents are operating across borders. A czar answerable only to political winds doesn't provide that. A framework does.