US Says Chinese AI Companies Are Copying American Technology

Three major US government security agencies have accused Chinese artificial intelligence (AI) companies of systematically copying AI technology from American firms like OpenAI, Google, and Anthropic.

According to the agencies, companies such as DeepSeek, Moonshot, Alibaba, and Z.ai are using a technique called "extraction" — a method where researchers build a new AI system by studying how a powerful existing AI model responds to questions. Think of it like reverse-engineering: if you want to understand how someone built a complex machine, you watch what it does and try to recreate it yourself. While this technique is common in AI research worldwide, the US agencies say Chinese companies are using it at an unusually large scale and with specialized focus, suggesting it's not just a side project but the core of how they develop their technology.

This accusation isn't new. In February, Anthropic (the company behind the AI assistant Claude) already complained that Chinese competitors were illegally extracting Claude's capabilities. The issue has now reached the highest levels of US politics — an adviser to President Trump raised concerns about the Chinese company Moonshot doing the same thing.

China's government denies the allegations, saying the quality of Chinese AI is simply the result of good science and independent technological development. The dispute reflects growing tensions between the US and China over who leads in artificial intelligence — a technology expected to shape everything from healthcare to national defense in coming years.