Anthropic Alleges Alibaba Conducted Extensive AI Model Extraction Operation

US artificial intelligence company Anthropic has formally accused Alibaba, the Chinese technology and e-commerce conglomerate, of conducting an extensive and illicit extraction campaign targeting Anthropic’s Claude AI model. According to a letter dated June 10, 2026, and reviewed by Reuters, the operation represents the largest known attack of its kind against Anthropic.

The campaign, described by Anthropic as a “distillation” effort, involved training a less capable AI model on the outputs generated by a more advanced one. This technique allows the replication of sophisticated AI capabilities without direct access to the original model’s underlying architecture or data.

Anthropic reported that the campaign took place between April 22 and June 5, 2026. During this period, operators affiliated with Alibaba and its AI research lab, Alibaba Qwen, allegedly created nearly 25,000 fraudulent accounts to generate more than 28.8 million exchanges with the Claude AI platform.

The company warned that such distillation efforts could accelerate China’s ability to develop AI models with capabilities comparable to Anthropic’s advanced Mythos Preview system. The letter was addressed to US Senators Tim Scott and Elizabeth Warren, chair and ranking member of the Senate Banking Committee, respectively, ahead of a scheduled hearing on artificial intelligence.

This accusation follows a broader context of US government concerns regarding intellectual property theft by Chinese entities. In April 2026, the White House publicly accused China of engaging in industrial-scale theft of AI intellectual property from US laboratories.

Anthropic expressed support for US government initiatives aimed at combating such attacks, including partnerships with private sector AI companies focused on threat intelligence sharing and coordinated defensive measures.

Earlier in 2026, Anthropic had identified similar extraction campaigns by other Chinese AI startups. In February, it disclosed that DeepSeek, a Chinese AI startup known for its low-cost AI model that disrupted the market in January 2025, along with two other labs—Moonshot AI and MiniMax—had illicitly extracted capabilities from Claude. DeepSeek’s campaign involved over 150,000 exchanges, Moonshot AI’s over 3.4 million, and MiniMax’s over 13 million.

Anthropic noted that these campaigns were increasing in both intensity and sophistication, emphasizing the need for rapid and coordinated action among industry stakeholders, policymakers, and the global AI community.

In a related development, Alibaba was added to the Pentagon’s list of Chinese military companies in June 2026, a designation Alibaba is currently contesting. Meanwhile, the US Commerce Department has refrained from placing DeepSeek on a trade blacklist despite national security concerns, reportedly to avoid escalating tensions with Beijing.

Two days after Anthropic’s letter was sent, on June 12, 2026, the Commerce Department imposed restrictions on Anthropic’s latest Mythos and Fable AI models. Officials cited fears that these models could be exploited by military intelligence agencies in China and other countries of concern.

These developments underscore the intensifying geopolitical and technological rivalry between the US and China in the AI sector, with significant implications for intellectual property protection and national security.

For ongoing updates on AI and technology policy, follow relevant congressional hearings and government announcements.