A growing dispute over AI model copying and data ethics has put one of the leading US AI startups at the centre of a global controversy. Anthropic has accused Chinese AI companies DeepSeek, Moonshot AI and MiniMax of carrying out “industrial-scale” distillation attacks on its Claude models. In a post on X and a blog published on February 23, the company said the labs created more than 24,000 fraudulent accounts and generated over 16 million exchanges with Claude to extract its capabilities and improve their own systems.
Anthropic said the labs used a method known as distillation, where smaller or less capable models are trained on the outputs of a stronger one. While distillation is a legitimate and widely used technique, including by frontier AI labs to create cheaper and smaller versions of their own models, the company warned it can also be misused. Competitors, it said, can extract advanced capabilities in a fraction of the time and at a fraction of the cost compared to building them independently. Anthropic added that such attacks are increasing in scale and sophistication and that the window to stop them is narrowing. It called for coordinated action from industry players, policymakers and the global AI community.
The company also described “illicitly distilled models” as a national security risk. It argued that US firms build safeguards to prevent state and non-state actors from using AI for malicious cyber activities, but illegally distilled models are unlikely to retain these protections. Anthropic warned that foreign labs could feed these unprotected capabilities into military, intelligence and surveillance systems, enabling authoritarian governments to deploy frontier AI for offensive operations. Since Claude is not offered commercially in China, the company said the labs bypassed restrictions using “Hydra Cluster architectures,” or networks of fraudulent accounts across its API and third-party cloud platforms. It added that large-scale distillation still requires significant compute, strengthening the case for US chip export controls.
The pushback was swift. Elon Musk wrote on X that “Anthropic is guilty of stealing training data at a massive scale and has had to pay multi-billion-dollar settlements for their theft. This is just a fact.” In December last year, Anthropic settled a first-of-its-kind AI copyright lawsuit with authors for $1.5 billion after reportedly using around 500,000 books without compensation. Reports also said US Defence Secretary Pete Hegseth recently met Anthropic CEO Dario Amodei at the Pentagon to discuss military use of Claude. On the same day as the accusations, a news outlet reported the meeting was not routine. Earlier this year, media reports claimed Claude was used in an operation leading to the capture of Venezuelan President Nicolás Maduro, though Anthropic has not confirmed this.
Also read: Viksit Workforce for a Viksit Bharat
Do Follow: The Mainstream LinkedIn | The Mainstream Facebook | The Mainstream Youtube | The Mainstream Twitter
About us:
The Mainstream is a premier platform delivering the latest updates and informed perspectives across the technology business and cyber landscape. Built on research-driven, thought leadership and original intellectual property, The Mainstream also curates summits & conferences that convene decision makers to explore how technology reshapes industries and leadership. With a growing presence in India and globally across the Middle East, Africa, ASEAN, the USA, the UK and Australia, The Mainstream carries a vision to bring the latest happenings and insights to 8.2 billion people and to place technology at the centre of conversation for leaders navigating the future.
