Washington accuses Chinese AI firms of ‘systematic’ theft from US rivals

In a recent advisory, Washington accused leading Chinese artificial intelligence firms, including DeepSeek and Moonshot AI, of systematically stealing capabilities from American AI models. The United States Cyber Defense Agency issued the warning ahead of an anticipated meeting between President Donald Trump and Chinese President Xi Jinping at the White House, where AI could be a key topic of discussion.

The accusations follow several previous claims of Chinese AI theft, which the US government and American AI companies have leveled against Beijing.

According to the advisory, Chinese AI companies are engaged in systematic extraction of proprietary functionalities and capabilities of US AI models through industrial-scale knowledge distillation campaigns. Distillation is a process where an AI model is trained to replicate the output of a more powerful model. While this technique is commonly used in the industry, Washington claims that Chinese firms have scaled it up to illegally replicate American systems.

The advisory highlights how Chinese companies like DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI allegedly used millions of exchanges and requests to pilfer the capabilities of US models from leading firms such as OpenAI, Anthropic, Google, and xAI. Although StepFun was not immediately reachable for comment, AFP sought statements from the other Chinese entities involved. Moonshot AI declined to comment on the allegations.

These actions, likely with Chinese government knowledge and approval, have been ongoing since at least late 2024, according to the advisory. With US AI models like those from Anthropic or OpenAI often preferred by companies over cheaper Chinese alternatives, President Trump faces pressure to respond to the growing dominance of Chinese AI technology.

In July, a group of AI startups wrote a letter to the Trump administration, urging caution against any outright bans or strict restrictions on foreign-made AI models, suggesting such measures should only be considered for government use. In the same month, Beijing retaliated by accusing US AI companies of using Chinese examples to train their models.