Open Source or Controlled Access? The New U.S.-China AI Divide

Podcast: The US-China Podcast · National Committee on U.S.-China Relations

Opublikowano: 2026-08-26

On June 12, 2026, the U.S. government placed Anthropic's two most capable models under export controls, the first time a frontier AI model itself, rather than the chips beneath, it was treated as a controlled item. While the controls were later removed, it presented a new source of uncertainty for closed models coming out of American frontier labs. At almost exactly the same time, Chinese firms, specifically Z.ai and Moonshot, released several highly capable open-source models, which sharpened a core tension in U.S. strategy: promoting open diffusion to extend American influence while locking down its most advanced, closed...