US Government Accuses Chinese AI Lab of Large‑Scale Model Theft

Anthropic sent a June 10, 2026 letter to the U.S. Senate Banking Committee alleging that the research division behind Alibaba’s Qwen large‑language model has carried out extensive intellectual‑property theft. The complaint cites a U.S. Office of Science and Technology Policy memorandum stating, “The United States government has information indicating that foreign entities, principally based in China, are engaged in deliberate, industrial‑scale campaigns to distill U.S. frontier AI systems.”

According to the letter, the Chinese effort uses tens of thousands of proxy accounts and jail‑breaking techniques to extract capabilities from American AI models, reproducing comparable performance at a fraction of the cost while stripping security and neutrality safeguards. Anthropic argues that such adversarial distillation exploits billions of dollars of U.S. AI research investment, threatens the security of AI systems, and could undermine America’s leadership in the frontier‑AI race.