China News

Alibaba's Claude Heist Accusation

AI model extraction allegation against Alibaba

Anthropic has accused Alibaba-linked operators of running a giant campaign to strip capabilities from Claude, turning a technical AI dispute into a fresh U.S.-China flashpoint.

According to The Wall Street Journal, which said it obtained a June 10 letter sent by Anthropic to U.S. senators Tim Scott and Elizabeth Warren, the company alleged that Alibaba and its Qwen AI unit carried out its largest known "distillation attack" to date. Business Standard, citing Bloomberg, reported the same core claim: nearly 25,000 fraudulent accounts were allegedly used to generate 28.8 million Claude exchanges.

Anthropic's accusation is explosive because distillation can be legitimate when a company trains smaller versions of its own models. The alleged abuse is different: using another lab's output at scale to copy valuable reasoning, coding, and agentic abilities without paying the full development cost.

The company has not publicly released the June 10 letter in the sources reviewed for this article. That matters. The numbers, the Alibaba link, and the Qwen connection remain Anthropic's allegations as reported by major outlets, not findings by a court or regulator.

What is confirmed is the broader fight. Anthropic publicly warned in February that Chinese AI labs DeepSeek, Moonshot, and MiniMax had used fraudulent accounts for more than 16 million Claude exchanges in earlier campaigns. It said those efforts targeted coding, agentic reasoning, tool use, and other high-value capabilities.

The Alibaba allegation raises the stakes because it names one of China's most important technology groups, not a smaller AI startup. It also lands while Washington is already treating frontier AI as a national-security asset and tightening controls on Chinese access to advanced chips and models.

Alibaba's detailed response was not found in the current reports reviewed here. Until it appears, the key unresolved question is simple: did Anthropic catch a massive AI copy job, or is this another accusation in an increasingly political model war?

Leaving Your Job? Read This First!!

Free legal consultation

Sources: The Wall Street Journal · Business Standard/Bloomberg · Anthropic