Anthropic Accuses Alibaba of Stealing Claude AI Capabilities
The scale of the alleged attack-more than 28.8 million interactions-far exceeds previous known distillation campaigns, and comes just as Alibaba released a model it claims matches Claude Opus 4.6's performance.
Reporting from 1 sources: GIGAZINE.
Anthropic has accused Alibaba's Qwen lab of conducting a massive distillation attack against its Claude AI model, involving over 28.8 million interactions via 25,000 fake accounts between April and June 2026. The company filed a complaint with White House officials and senators, calling for joint government-industry action against illegal distillation.
Anthropic has sent a complaint to White House officials and senators accusing Alibaba's Qwen lab of running a large-scale distillation attack against the Claude AI model. The company says the campaign involved over 28.8 million interactions with Claude using approximately 25,000 fake accounts between April and June 2026. That volume surpasses earlier distillation attacks by Chinese AI ventures DeepSeek, MiniMax, and Moonshot AI, which together totaled 16 million interactions via 24,000 fake accounts before February 2026.
Distillation extracts inputs and outputs from a high-performance model to improve other models. Anthropic is cooperating with OpenAI and Google to counter such attacks. In June 2026, Alibaba released Qwen3.7-Plus, which it claims performs comparably to Anthropic's Claude Opus 4.6. An Alibaba spokesperson declined to comment on the accusation.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.