China's AI Heist (www.foreignaffairs.com)

Recent developments in the U.S.-China AI competition spotlight the emergence of open-weight, local AI models that can run efficiently on consumer hardware, such as laptops and smartphones. This shift allows users to leverage powerful AI capabilities without the need for cloud services, enhancing accessibility and control. However, a significant concern arises as many of the most advanced open-weight models are being developed by Chinese companies, which capitalize on a practice known as distillation. This involves training smaller models to replicate the outputs of larger ones, effectively extracting the capabilities of American AI models while U.S. firms face restrictions from contractual obligations that prevent similar practices. The implications extend beyond mere market competition; they signal a potential geopolitical dependence on Chinese AI technology. As developers gravitate toward local models to avoid cloud costs and data privacy concerns, they risk favoring Chinese infrastructure and chips over U.S. alternatives, thus deepening dependency on China's AI ecosystem. With the overlapping risks of technological influence and security vulnerabilities, U.S. policymakers face the challenge of fortifying domestic AI capability while counteracting unauthorized Chinese practices. Effective strategies to promote U.S. leadership in open-weight AI while safeguarding against distillation risks will be crucial as the global landscape shifts toward localized AI solutions.