AI distillation becomes a flashpoint in U.S.-China tech rivalry

AI distillation, a technique that uses the outputs of a larger model to train a smaller one, has become a major topic in the tech industry and in Washington. The debate intensified after Chinese lab Moonshot AI released Kimi K3, which some users said was competitive with leading U.S. models. U.S. officials and some industry figures argue that distillation can amount to unauthorized use of proprietary model outputs, while others say it is a common and legitimate way to make AI systems cheaper and more efficient. The issue has raised concerns about intellectual property, competition, and national security. In response to the backlash, Nvidia, Microsoft, Meta, Palantir, and more than 20 other companies signed a letter urging policymakers not to impose premature limits on openweight AI models. Anthropic, meanwhile, says its models have been distilled at scale and has described the practice as a security concern.