White House officials have accused China of running industrial‑scale campaigns to siphon American AI research, citing a memo from Science and Technology Policy Director Michael Kratsios that points to deep‑seeking distillation attacks.

The memo warns that foreign actors, mainly based in China, use thousands of proxy accounts and jailbreaking techniques to harvest outputs from U.S. frontier models. Kratsios said the administration would share intel with U.S. AI firms and pursue accountability measures.

China’s embassy dismissed the claims as "pure slander," while U.S. AI firms such as Anthropic and OpenAI have already flagged distillation by DeepSeek, Moonshot and MiniMax as a national‑security risk. The House Foreign Affairs Committee has introduced bills that could place distilling groups on the entity list, tightening export controls.

The dispute arrives ahead of a Trump‑Xi summit and heightens tensions around AI dominance. U.S. companies face mounting pressure to protect IP while navigating export‑control constraints, potentially reshaping the competitive edge in the global AI market.