Anthropic Accuses Chinese AI Firms of Stealing Claude Model's Intelligence
AI firm Anthropic has accused Chinese companies DeepSeek, Moonshot, and MiniMax of using "distillation" to illegally copy its Claude chatbot's capabilities. Anthropic claims over 16 million queries from 24,000 fake accounts were used to train smaller models, bypassing significant development costs and representing potential intellectual property theft.
Key points
- AI developer Anthropic has accused Chinese firms DeepSeek, Moonshot, and MiniMax of intellectual property theft.
- The companies allegedly used a technique called "distillation" to extract intelligence from Anthropic's Claude chatbot.
- Anthropic claims over 16 million exchanges were made using 24,000 fraudulent accounts to train competitor models.
- This method allows smaller models to mimic larger ones without incurring the original massive training costs.
- Anthropic views this industrial-scale operation as a parasitic economy and potential IP theft.
Artificial intelligence company Anthropic has publicly accused three Chinese AI firms—DeepSeek, Moonshot, and MiniMax—of illicitly siphoning the capabilities of its flagship Claude chatbot. Anthropic alleges that these companies engaged in an "industrial-scale" operation using a technique known as model distillation.
This process involves repeatedly querying a powerful AI model, referred to as the "teacher," and using the resulting high-quality outputs to train a smaller, less resource-intensive "student" model. Anthropic claims that over 16 million interactions with Claude were generated through approximately 24,000 fraudulent accounts. The company asserts that this method allows competitors to bypass the substantial costs associated with developing advanced AI models and constitutes a form of intellectual property theft.
Model distillation, while not inherently illegal, becomes problematic when used by competitors to replicate proprietary AI capabilities without authorization or investment in original research and development. The alleged use of fake accounts and massive query volumes by the Chinese labs has brought this practice to the forefront of industry concerns regarding AI development ethics and IP protection.
Sources
The WireByte editorial team synthesises technology news from multiple primary sources, verifies the facts, and links every source. Articles are produced with AI assistance and reviewed under our editorial policy.