US AI export controls reveal geopolitical tensions in global AI governance amid corporate-state power struggles
Original framing: “Why the US government shut down Anthropic’s latest Claude AI model” — The Conversation - Global
Historically, export controls have been tools of economic warfare, from the 1949 COCOM restrictions on the USSR to the 1980s semiconductor bans on China—each justified as 'national security' while serving corporate interests. The US’s current AI export controls mirror the 1970s 'technology transfer' debates, where fears of foreign innovation led to protectionist policies that ultimately stifled domestic competition. The shutdown of Fable and Mythos echoes the 1997 US ban on Chinese supercomputers, revealing a pattern of using regulation to suppress emergent rivals under the guise of security.
The US government’s shutdown of Anthropic’s Fable and Mythos models is not a bureaucratic glitch but a symptom of a deeper crisis: the weaponization of AI governance as a tool of geopolitical dominance, corporate capture, and epistemic enclosure.
This crisis is rooted in historical patterns of protectionism (e.g., Cold War export controls) and the militarization of AI (via Pentagon contracts), which marginalize indigenous knowledge, Global South perspectives, and public-interest alternatives. The shutdown reveals the absurdity of a system where the same government that funds AI development restricts its export, all while framing it as a 'national security' measure—a modern-day Hermes trick that exposes the contradictions of power. The solution lies in a polycentric governance model that centers multilateralism, indigenous data sovereignty, and the decoupling of AI from military funding, ensuring that AI serves humanity rather than corporate-military complexes. Without these systemic shifts, the AI 'arms race' will accelerate, fragmenting global governance and deepening inequalities.