Anthropic accuses Chinese AI labs of mining Claude as US debates AI chip exports
Anthropic is publicly accusing DeepSeek, Moonshot AI, and MiniMax of creating over 24,000 fake accounts to extract capabilities from Claude through more than 16 million exchanges — specifically targeting its strongest areas: agentic reasoning, coding, and tool use. The timing is pointed: the accusations land right as Washington debates loosening AI chip exports to China, and Anthropic is explicitly arguing that distillation at this scale requires advanced chips, reinforcing the case for export controls. Beyond the competitive threat, Anthropic warns that distilled models likely strip out safety guardrails, meaning dangerous capabilities could proliferate without any of the protections built into the originals.