Anthropic has charged Alibaba with conducting the largest distillation campaign against Claude.

      TL;DR: Anthropic has accused Alibaba’s Qwen lab of executing a large-scale distillation campaign against a U.S. AI company, claiming that operators associated with Qwen used approximately 25,000 fake accounts to conduct nearly 29 million exchanges with Claude from April to June. This allegation represents the first time Anthropic has identified a major Chinese tech firm as the perpetrator of a distillation attack, following earlier allegations against smaller Chinese AI startups. Distillation involves feeding a leading AI model targeted queries to harvest its responses for training a competing system. The White House has expressed concerns about this practice. Anthropic's warnings highlight that such methods enable Chinese labs to replicate advanced AI models at a lower cost, often without necessary safety measures. Concurrently, lawmakers are proposing amendments to defense legislation to blacklist or sanction Chinese firms accessing U.S. AI model outputs improperly. At the same time, Anthropic is navigating its own challenges with the U.S. government regarding export controls on its models. The outcome of these developments may significantly influence the regulatory landscape for U.S. AI firms and the competition surrounding intellectual property protections for AI systems.