Anthropic Accuses Alibaba Of Illicit Access To Claude AI Models — Notifies White House

U.S. artificial intelligence developer Anthropic has accused Chinese technology giant Alibaba of a coordinated, industrial-scale operation to bypass geo-restrictions and illicitly scrape data from its proprietary AI model, Claude.

• Anthropic alleges that operators linked to Alibaba's Qwen AI lab utilized thousands of fake accounts to bypass safety barriers.
• The systematic extraction utilized a process that enables foreign competitors to build rival software at a fraction of the market cost, which Anthropic warns risks creating copycat models.
• The startup has officially detailed the breach to White House officials and U.S. senators, urging Washington to enact tighter regulatory safeguards.

Claude maker Anthropic officially accused Chinese technology and e-commerce giant Alibaba Group Holding on Wednesday of executing a sweeping, highly coordinated operation to systematically siphon the core capabilities of its proprietary AI system.

In a letter sent to U.S. senators and White House officials, the San Francisco-based developer detailed what it characterized as the largest known data-extraction campaign launched by a Chinese entity against its technology to date, Bloomberg exclusively reported.

According to Anthropic, operators connected to Alibaba’s specialized Qwen AI laboratory deployed thousands of fraudulent user accounts to evade restrictions and access its Claude chatbot models. The unauthorized campaign directly circumvents Anthropic's strict geographic distribution rules, which explicitly prohibit its software from being deployed or accessed inside of China.

Alibaba (BABA) stock dropped nearly 3% on Wednesday.

Anthropic’s Allegations on Alibaba: Details

Anthropic claimed Alibaba is using a method called "adversarial distillation." Through this mechanism, outside entities repeatedly prompt an advanced model to harvest its reasoning patterns and data structure. Competitors can subsequently use those responses to train their own internal AI software, bypassing millions of dollars in primary research and development costs.

“These distillation attacks are carried out illicitly, systematically, and at an industrial scale to harvest US AI capabilities across frontier labs and repackage them as their own without incurring the training and R&D costs required to train US frontier models,” Anthropic wrote in its letter, accessed by Bloomberg. 

Anthropic noted that when high-tier AI models are cloned through unauthorized distillation, the resulting software typically strips away the foundational guardrails and safety protocols engineered into the original U.S. platforms.

Anthropic urged federal lawmakers to implement more stringent enforcement and defensive frameworks to prevent foreign tech labs from systematically stripping intellectual property from domestic AI research firms.

BABA Stock: Retail View 

Retail sentiment on Stocktwits was ‘bullish’ with ‘high’ message volumes. 

“Nobody wants Chinese equities right now - and that’s usually where the real opportunity starts forming,” said one user. 

BABA stock has lost 32.4% year-to-date. 

For updates and corrections, email newsroom[at]stocktwits[dot]com.<