Anthropic has publicly accused Alibaba of illicitly extracting the capabilities of its Claude AI model through a large-scale distillation attack.
The letter, addressed to Senators Tim Scott and Elizabeth Warren, claims that Alibaba-linked operators performed almost 29 million exchanges with Claude, using thousands of fabricated accounts. According to the US-based company, the operation targeted Claude’s most advanced skills – long‑form reasoning, complex problem solving and decision‑making approaches – thereby enriching Alibaba’s own models.
Anthropic warned that such activity transforms "hundreds of billions of dollars" of American R&D into a subsidy for geopolitical competitors. The letter cites Department of Defense findings linking Alibaba to the Chinese military and highlights similar tactics used by other Chinese firms such as BYD and Baidu.
Alibaba has denied the allegations and has filed a lawsuit to remove its name from the Pentagon blacklist. Industry experts say the incident underscores the need for tighter AI safeguards and international cooperation.
As Anthropic prepares for a high‑profile IPO, the security concerns around its advanced models – like Mythos – loom large, especially given the questions about malicious use for system‑weakening attacks.
