The Great AI Debate: Sovereignty, Regulation, and Distributed Power

A disagreement on X between Anthropic CEO Dario Amodei, investor Gavin Baker, and Anthropic researcher Sholto Douglas turned into a much bigger argument about where power should sit in an AI-driven world.Gavin’s concern is that regulating increasingly powerful AI could eventually concentrate control inside a small group of frontier labs and governments. His preferred direction is broader distribution: more models, more competitors, more open weights, and fewer opportunities for any single company to dominate advanced intelligence.Dario argues that this misses part of the problem. Frontier AI may naturally concentrate power because the best systems increasingly depend on enormous amounts of compute, chips, electricity, data centers, and capital. Open models can broaden access, but they do not automatically decentralize the infrastructure required to build and operate frontier AI.The conversation also gets into regulatory capture, biological and cybersecurity risks, the role of government testing, open-weight AI, and Dario’s concern that extreme economic concentration could eventually weaken the political and economic leverage ordinary people have today.EP 63 breaks down both sides without reducing this to another open-versus-closed AI argument.The bigger question is becoming much harder to ignore:If intelligence becomes one of the world’s most valuable resources, who gets to control it?