Anthropic has told the Senate Banking Committee that operators tied to Alibaba’s Qwen lab ran roughly 25,000 fraudulent Claude accounts through 28.8 million exchanges between April 22 and June 5, in what the company calls the largest known adversarial distillation campaign against a US AI lab. The June 10 letter, addressed to chair Tim Scott and ranking member Elizabeth Warren and made public by CNBC on June 24, arrives with the political scaffolding already halfway built.
The scale is the point, and also isn’t. In February, Anthropic accused DeepSeek, Moonshot AI, and MiniMax of running a joint campaign involving roughly 24,000 fake accounts and more than 16 million Claude exchanges. Four months later the numbers have nearly doubled. Sarah Heck, Anthropic’s policy head, framed the pattern in the letter as an effort to work “illicitly, systematically, and at industrial scale to harvest U.S. AI capabilities across frontier labs and repackage them as their own without incurring the training and R&D costs required to train U.S. frontier models.”
According to The Next Web, which viewed the letter via Bloomberg, the queries concentrated on software engineering and agentic reasoning, Claude’s most commercially valuable skills. That’s not indiscriminate scraping. It’s targeted extraction of the workloads a $965 billion company built its enterprise pricing around, ahead of a confidentially filed IPO expected this autumn on top of last year’s $65 billion Series H.
The policy machinery is already moving. Senators Bill Hagerty and Andy Kim plan to attach an amendment to must-pass defense legislation, and Representatives Bill Huizenga and Sydney Kamlager-Dove have a bipartisan House bill targeting Chinese firms that improperly access US model outputs. An April OSTP memorandum signed by director Michael Kratsios already commits the government to sharing foreign-distillation intelligence with US labs. Alibaba was added to the Pentagon’s Chinese military companies blacklist on June 8 and is suing to get off it.
Domestic access is tightening on the same axis. Earlier this month, the Trump administration ordered Anthropic to suspend access to Fable 5 and Mythos 5 for any foreign national worldwide, including its own foreign-national employees, citing unspecified national security authorities. Commerce Secretary Howard Lutnick signed the order; Anthropic complied.
Alibaba didn’t respond to requests for comment, and none of Anthropic’s claims have been independently confirmed. But independent confirmation isn’t what the letter is optimized for. It’s a filing designed to convert an adversarial-usage log into statute before the IPO roadshow, using a committee that oversees capital markets rather than one that oversees technology. The venue tells you what kind of fight Anthropic thinks this actually is.
Sources
- https://www.cnbc.com/2026/06/24/anthropic-alibaba-distillation-campaign.html
- https://thenextweb.com/news/anthropic-accuses-alibaba-distillation-claude-qwen
- https://www.techradar.com/ai-platforms-assistants/anthropic-accuses-alibaba-of-copying-claude-by-asking-it-millions-of-questions-and-sets-the-stage-for-a-new-ai-war
- https://www.tomshardware.com/tech-industry/artificial-intelligence/anthropic-claims-that-chinas-alibaba-illicitly-distilled-its-models-from-april-to-june-2026-says-effort-involved-25-000-fake-accounts-and-28-8-million-exchanges-on-claude
- https://www.pymnts.com/news/artificial-intelligence/2026/anthropic-accuses-alibaba-of-running-29-million-fake-queries-to-clone-claude/