Anthropic accuses Chinese AI labs of mining Claude as US debates AI chip exports

image via TechCrunch

February 23, 2026, 7:57 PM

• •Anthropic alleges that Chinese AI companies used distillation to copy capabilities from its Claude AI model.
• •The accused firms are DeepSeek, Moonshot AI, and MiniMax.
• •These attacks targeted Claude's agentic reasoning, tool use, and coding capabilities.
• •The scale of the attacks has reignited the debate over export controls on AI chips.
• •Anthropic warns of national security risks if safeguards against AI misuse are compromised.

Anthropic, an AI research company, has accused three Chinese AI firms (DeepSeek, Moonshot AI, and MiniMax) of using over 24,000 fake accounts to extract data from its Claude AI model through a technique called distillation. These firms allegedly targeted Claude's advanced reasoning, tool use, and coding capabilities, amassing millions of exchanges to improve their own AI models. The accusations come amidst discussions about enforcing export controls on AI chips, as the scale of these distillation attacks necessitates access to advanced computing power. Anthropic emphasizes that such attacks pose national security risks by potentially undermining safeguards against misuse of AI, and calls for industry-wide and governmental responses.

Read original article