China says AI distillation allegations are "misguided"

China has responded to recent allegations from U.S. artificial intelligence (AI) firms, including Anthropic, that Chinese labs are engaging in large-scale "distillation" attacks to extract capabilities from American AI models. The Chinese government has described these claims as "misguided," emphasizing that AI development is a global endeavor and that risk undermining international collaboration.

Anthropic has reported that Chinese AI labs, including Alibaba's Qwen AI, DeepSeek, Moonshot AI, and MiniMax, have used tens of thousands of fraudulent accounts to generate millions of interactions with its Claude models. These interactions, the company argues, are used to train Chinese models through distillation—a technique that replicates advanced model behavior. Anthropic has attributed these activities to bypass U.S. export controls.

The U.S. government has already imposed export restrictions on Anthropic’s most advanced models, citing national security concerns. Anthropic has urged lawmakers to treat distillation attacks as a national security issue, calling for stronger enforcement and sanctions.

China, however, has rejected these allegations, asserting that AI innovation should not be restricted by geopolitical tensions. The country has emphasized its commitment to ethical AI development and has called for balanced approach to regulation.

As the U.S. weighs further policy responses, the debate over distillation highlights the growing tension between national security interests and the open exchange of AI technologies. The outcome of this dispute could shape the future of AI governance and influence how intellectual property is protected in the rapidly evolving frontier of artificial intelligence.