White House splits AI policy: open weights, China crackdown

The Trump administration is drawing a new line in AI policy: defend open-weight models, target Chinese companies. Axios reports the White House accused China's Moonshot of using distillation to extract U.S. technology while simultaneously defending open-weight models that businesses and researchers rely on. This bifurcated approach matters because it could allow U.S. action against Chinese AI companies without broadly restricting the open-weight ecosystem that drives competition in the model layer. The policy distinction creates a structural advantage for U.S. providers who can operate under open-weight frameworks while Chinese competitors face restrictions. The economics shift accordingly: U.S. model providers get regulatory cover to maintain open-weight strategies that commoditize inference costs, while Chinese companies get cut off from the same open-source training techniques that drive cost compression across the industry. This policy split effectively creates two parallel AI infrastructure markets with different cost curves and innovation dynamics. The immediate impact will be on how Chinese companies source their model development, potentially accelerating their move toward fully proprietary stacks that cannot leverage global open-source improvements.