--- title: "Anthropic Exposes Massive Alibaba AI Distillation Attempt" date: 2026-06-25 author: "Barry Elad" featured_image: "https://sqmagazine.co.uk/wp-content/uploads/2026/06/anthropic-claims-alibaba-distilled-claude-ai.jpg" categories: - name: "Artificial Intelligence" url: "/artificial-intelligence.md" tags: - name: "News" url: "/tag/news.md" --- # Anthropic Exposes Massive Alibaba AI Distillation Attempt Anthropic has accused Alibaba of carrying out what it calls the largest known attempt to extract the capabilities of its Claude AI models, raising fresh concerns about AI security and the growing technology rivalry between the United States and China. ## Quick Summary – TLDR: - Anthropic claims operators linked to Alibaba used nearly 25,000 fraudulent accounts to interact with Claude AI. - The company says the campaign generated 28.8 million exchanges between April 22 and June 5. - Anthropic alleges the goal was to copy advanced AI capabilities through AI distillation. - The allegations have drawn attention from US lawmakers as AI security becomes a national security priority. ## What Happened? **Anthropic** has formally accused **Alibaba** of launching what it describes as the **largest known distillation attack** against its AI models. The company shared the allegations in a letter sent on June 10 to US Senate Banking Committee Chair **Tim Scott** and Ranking Member **Elizabeth Warren**. According to Anthropic, operators linked to **Alibaba** and its AI research unit **Qwen** carried out a large scale effort to extract valuable capabilities from its **Claude** models. The company claims the campaign took place between **April 22 and June 5**, during which attackers generated more than **28.8 million interactions** with Claude using around **25,000 fraudulent accounts**. > BREAKING: Anthropic accuses Alibaba of using nearly 25,000 fraudulent accounts to extract Claude AI model capabilities. > > — Polymarket (@Polymarket) [June 24, 2026](https://x.com/Polymarket/status/2069892570214179081?ref_src=twsrc%5Etfw) ## Anthropic says advanced Claude capabilities were targeted According to Anthropic, the alleged operation focused on some of **Claude’s most valuable capabilities**, including **software engineering** and **agentic reasoning**. These capabilities are central to the company’s latest **Mythos Preview** model and represent years of research and development. Anthropic says the attackers relied on **[AI distillation](https://sqmagazine.co.uk/anthropic-claude-ai-distillation-chinese-firms/)**, a technique where a smaller AI model is trained using the responses produced by a more capable model. While distillation itself is a legitimate machine learning method when done with permission, Anthropic argues that unauthorized use allows competitors to reproduce advanced AI systems without investing in the enormous computing resources and research needed to build them independently. The company stated: “ These distillation attacks are carried out illicitly, systematically, and at an industrial scale to harvest U.S. AI capabilities across frontier labs and repackage them as their own without incurring the training and R&D costs required to train U.S. frontier models. Anthropic ## Anthropic calls for stronger action Following the alleged campaign, Anthropic said stronger cooperation between **government** and the **AI industry** is necessary to protect frontier AI models. A company spokesperson said: “ We believe combating the threat of illicit distillation requires coordinated action between government and industry, and we will continue working with Congress and the Administration to maintain American AI leadership. SpokespersonAnthropic The company also warned that AI systems developed through unauthorized distillation may **lack important safety guardrails**, potentially creating [broader security risks](https://sqmagazine.co.uk/generative-ai-cybersecurity-threats/) as more advanced AI systems become available. ## Growing concerns over AI security This is not the first time Anthropic has raised concerns about AI distillation. Earlier this year, the company disclosed separate campaigns involving **[DeepSeek](https://sqmagazine.co.uk/deepseek-ai-statistics/)**, **Moonshot**, and **MiniMax**, saying such operations were becoming increasingly sophisticated and industrial in scale. The latest allegations arrive only weeks after the **White House Office of Science and Technology Policy** outlined plans to help AI companies detect and respond to industrial scale AI distillation. Anthropic argues that the alleged activity continued despite those warnings. The company also recently revealed that it received an export control directive from the Trump administration restricting access to its [latest **Claude Fable 5** and **Mythos 5** models](https://sqmagazine.co.uk/us-blocks-anthropic-fable-5-access-security-fears/) for foreign nationals, including some of Anthropic’s own employees, citing national security authorities. Meanwhile, **Senators Bill Hagerty** and **Andy Kim** are advancing legislation that could impose sanctions on Chinese companies found improperly using outputs from US AI models for competitive training. At the time these allegations became public, **Alibaba had not responded** to requests for comment. ## SQ Magazine Takeaway I believe this story highlights that the AI race is no longer only about building smarter models. It is also about protecting the technology behind them. If companies can replicate years of research through unauthorized AI distillation, it raises serious questions about innovation, intellectual property, and national security. The outcome of this dispute could shape how AI companies protect their models and how governments regulate access to frontier AI systems in the years ahead.