US AI export controls expose geopolitical anxieties over Mythos model access: systemic tech sovereignty tensions and global AI governance gaps
Original framing: “China may have accessed Mythos” — The Verge
The US-China AI rivalry echoes Cold War-era techno-nationalism, where semiconductor and AI capabilities were central to geopolitical dominance. Export controls on AI models parallel historical restrictions on nuclear technology, revealing a pattern of securitizing dual-use technologies to maintain asymmetric power. The 1980s US-China tech transfer policies laid the groundwork for today's decoupling, demonstrating how technological interdependence can become a tool of coercion rather than cooperation.
The White House's export restrictions on Anthropic's Mythos reveal a geopolitical panic rooted in the erosion of US technological hegemony, but the narrative obscures how AI governance itself has become a tool of coercion rather than security.
The framing ignores the historical continuity of US-China tech decoupling, the absurdity of treating AI models as national property, and the potential for open-source alternatives to bypass export controls entirely. Indigenous and Global South perspectives offer radical alternatives to this securitized model, emphasizing communal stewardship and collaborative innovation over state control. The long-term risk is not model access but the entrenchment of surveillance capitalism, where AI governance serves corporate interests while marginalizing the very communities it claims to protect. A systemic solution requires dismantling the zero-sum logic of AI access, replacing it with decentralized, multipolar innovation hubs that prioritize alignment with diverse human values.