US accuses China of industrial-scale AI model distillation, will share Intel (thenextweb.com)

The White House Office of Science and Technology Policy (OSTP) has released a memo accusing China of conducting "industrial-scale" AI model distillation, which involves using large volumes of queries to illicitly train cheaper AI models that mimic American technology. This move is significant as it underscores the U.S. government's commitment to protecting AI innovations deemed crucial for national security. The memo comes just weeks before a high-profile summit between President Trump and President Xi, positioning AI protection as both a geopolitical concern and a bargaining tool. The allegations are backed by significant evidence from U.S. companies, including OpenAI and Anthropic, which have documented attempts by entities like DeepSeek to circumvent access restrictions and extract proprietary outputs through fraudulent accounts. The memo encourages intelligence sharing among U.S. companies regarding foreign distillation attacks and explores accountability measures, but it does not stipulate immediate sanctions or regulations. As AI distillation poses a new frontier challenge beyond traditional theft, the U.S. aims to establish a robust defense against it, signaling an urgent need for a legal framework to navigate the complexities of AI protection and intellectual property rights.