Anthropic Accuses Chinese Firms of AI Theft

Anthropic Alleges Chinese AI Firms Stole Claude Capabilities

WASHINGTON: (Web Desk) – US artificial intelligence firm Anthropic said it has uncovered large-scale efforts by three Chinese AI companies to extract capabilities from its Claude chatbot without authorization, describing the activity as industrial-level intellectual property theft. According to the company, DeepSeek, Moonshot AI, and MiniMax relied on a method known as “distillation,” which involves using responses from a more advanced AI model to enhance the performance of a smaller or less capable system.

Anthropic said the campaigns have become increasingly sophisticated and urgent, warning that time to respond is limited. While distillation is commonly used within companies to develop more efficient versions of their own AI systems, Anthropic alleges these firms used the method improperly to replicate Claude’s strengths.

The company claims the three labs carried out roughly 16 million interactions with Claude through more than 24,000 fraudulent accounts. MiniMax reportedly led the largest effort, generating over 13 million exchanges. The operations focused heavily on coding, advanced reasoning, and tool usage — areas where Claude is considered particularly strong.

Pakistan Secure Second Consecutive Win Over USA in T20 World Cup

Anthropic also alleged the companies bypassed restrictions on commercial access from China by routing traffic through proxy services that managed networks of fake accounts. It warned that AI models developed through unauthorized distillation may lack important safety guardrails, including protections against misuse in areas such as bioweapons development or cyberattacks.

Earlier this month, OpenAI, the creator of ChatGPT, raised similar concerns with US lawmakers, accusing Chinese firms of attempting to leverage American AI innovations without proper authorization. Anthropic has now called for coordinated action between industry leaders and governments to address what it described as a growing national security risk that no single company can solve alone.

Comments are closed, but trackbacks and pingbacks are open.