Anthropic released a threat intelligence report accusing multiple Chinese AI labs of unauthorized use of its Claude model's output to train their own models through identity concealment, fraud, credential theft, and other means, defining such actions as 'illegal distillation.' The report revealed that between December 2025 and August 2026, Anthropic blocked multiple unauthorized actions from Chinese AI labs, with Alibaba-linked activities being the largest in scale, totaling over 151 million interactions from May to July 2026. Additionally, Moonshot AI and DeepSeek were also accused of similar practices, forwarding user interactions to Claude without consent to extract reasoning texts for training their own models. In response, Anthropic CEO Dario Amodei called on U.S. regulators to crack down on distillation and urged the government to ban the sale of high-performance AI chips to China. However, Y Combinator President Garry Tan publicly opposed regulation, arguing that manufacturers have no right to dictate how customers use model outputs and that the U.S. should establish its own distillation framework to keep access to intelligence closer to a public good.