White House says China running industrial-scale AI distillation campaigns

The White House has accused China of running “industrial-scale” AI model distillation campaigns, prompting a new policy memo committing the US government to share threat intelligence with OpenAI, Anthropic, and Google. Evidence from February shows Chinese labs like DeepSeek, MiniMax, and Moonshot AI allegedly created 24,000 fraudulent accounts and generated over 16 million interactions with Claude to extract model behavior. The memo signals a shift toward model-level protection as hardware export controls prove insufficient, especially with smuggling schemes and China’s improving domestic chips. With Congress advancing the Deterring American AI Model Theft Act, the US is preparing a broader enforcement architecture ahead of the Trump–Xi summit.

Read the full story on TNW →