Anthropic accuses Alibaba of mass scale AI plagiarism

Advertisement

American artificial intelligence startup Anthropic has accused Chinese technology giant Alibaba Group Holding Ltd. of executing a massive, illicit campaign to access its advanced Claude AI model. According to a letter sent by Anthropic to several US senators and White House officials, operators linked to Alibaba’s Qwen AI lab used thousands of fraudulent accounts to bypass Anthropic’s restrictions keeping its products out of China. The startup stated that the operation specifically targeted Claude’s most valuable capabilities, such as software engineering and agentic reasoning, making it the largest known attempt by a Chinese firm to piggyback on the breakthroughs of a premier US AI lab.

In response to these findings, Washington lawmakers are moving quickly to penalize foreign exploitation of domestic AI. Senators Bill Hagerty (R-TN) and Andy Kim (D-NJ) plan to introduce an amendment to mandatory defense legislation that would blacklist or sanction Chinese companies caught improperly using US AI model outputs. A similar bipartisan measure is being pushed in the House by Representatives Bill Huizenga (R-MI) and Sydney Kamlager-Dove (D-CA). Anthropic noted that Alibaba’s campaign directly defied an April memo issued by Michael Kratsios, director of the White House Office of Science and Technology Policy under the Trump administration, which warned against the industrial scale exploitation of American models.

The economic and strategic implications of this data harvesting are massive for Silicon Valley, where unauthorized scraping is estimated to cost US labs billions of dollars. Anthropic, which is currently valued by private investors at $965 billion and preparing for an initial public offering, warned that this behavior threatens its business with cheap, imitation products that siphon away customers. In its appeal to the government, Anthropic urged US officials to implement strict penalties for these practices, enforce advanced chip export controls, and clarify antitrust guidelines so that competitive American labs can more freely share security data.

Advertisement

The technical scale of the alleged breach is unprecedented, involving 28.8 million exchanges with Claude between April and June through nearly 25,000 fraudulent accounts. Following the publication of these allegations, Alibaba’s American depositary receipts sank more than 3% to a session low of $99.10 in New York. Anthropic explained that Alibaba utilized a method known as ‘adversarial distillation’, a practice where a developer harvests the high-value outputs of a cutting-edge model to train a rival chatbot at a fraction of the original R&D cost. The startup warned that systems cloned through this method are heavily repackaged and typically lack vital safety guardrails.

While Alibaba offered no comment on the allegations, the situation has accelerated an alliance among top US AI developers. Anthropic, OpenAI, and Google have joined forces to share information regarding terms-of-service violations involving distillation, having previously flagged similar attempts by other Chinese startups like DeepSeek and Minimax. The accusations add to a volatile week for Alibaba, which was recently added to a US Defense Department blacklist for allegedly supporting China’s armed forces a designation Alibaba denies and is currently fighting via a lawsuit filed against the Pentagon.

Despite seeking government intervention against its Chinese rivals, Anthropic remains locked in its own high-stakes dispute with the Trump administration. The US Commerce Department recently slapped strict export controls on Anthropic’s top two models, Fable 5 and Mythos 5, citing severe national security concerns. Consequently, Anthropic was forced to disable access to these tools. Recent meetings between Anthropic’s senior technical staff and White House officials have yielded little progress toward resolving the standoff or restoring service to the restricted systems.

Also Read: PM Viksit Bharat Rozgar Yojana: Odisha Secures ₹37.96 Cr for Manufacturing Jobs

Advertisement