US security agencies have accused several leading Chinese AI companies of systematically extracting knowledge from American AI models and using it to develop their own systems. The agencies have warned US AI developers to strengthen protections around their models.
The National Security Agency, FBI and Cybersecurity and Infrastructure Security Agency said the companies have used AI distillation since at least 2024 to obtain information from US models “at an industrial scale”. The agencies specifically named DeepSeek, Moonshot AI, Alibaba Group Holding Ltd., MiniMax, StepFun and Z.AI Co.
The warning comes weeks before US President Donald Trump is expected to meet Chinese President Xi Jinping in Washington. The allegations could add further tension between the US and China as both countries compete for leadership in AI.
Distillation can be a legitimate method when an AI developer uses its own advanced model to train a more efficient system. However, using the technique without approval to extract knowledge from another company’s model is considered an improper shortcut.
The latest security alert detailed which US AI systems were allegedly targeted by each Chinese company and how the extracted information was used, including improving mathematical problem-solving and code review capabilities.
The agencies also said the companies are likely siphoning information from American models “with knowledge of the Chinese government”. They accused the firms of deliberately bypassing restrictions imposed by US AI companies.
“China-based AI companies route distillation requests through multiple pathways to gain unauthorized access, consequently violating U.S. AI companies’ terms of use,” the agencies said.
Chinese embassy spokesperson Liu Chang rejected the claims, calling them a “deliberate attack on China’s development and progress in the AI industry.” He added that “China opposes politicizing and instrumentalizing trade and tech issues. Such actions will only stifle global AI advances and serve no one’s interests.”
The security agencies urged US AI developers to take “immediate action”, including changing responses to suspected malicious distillation attempts and sharing intelligence on such campaigns.
OpenAI and Anthropic have also accused Chinese competitors of accessing their models without authorisation. In June, Anthropic accused Alibaba of using thousands of fraudulent accounts to “illicitly” access its Claude model.
The Trump administration has pledged further action, while US lawmakers are considering penalties for Chinese companies accused of unfairly copying American AI technology. Treasury Secretary Scott Bessent also warned in July that sanctions could be imposed over intellectual property theft. His comments followed Moonshot’s release of Kimi K3, which raised concerns over China’s growing AI capabilities.
Also read: Viksit Workforce for a Viksit Bharat
Do Follow: The Mainstream LinkedIn | The Mainstream Facebook | The Mainstream Youtube | The Mainstream Twitter
About us:
The Mainstream is a premier platform delivering the latest updates and informed perspectives across the technology business and cyber landscape. Built on research-driven, thought leadership and original intellectual property, The Mainstream also curates summits & conferences that convene decision makers to explore how technology reshapes industries and leadership. With a growing presence in India and globally across the Middle East, Africa, ASEAN, the USA, the UK and Australia, The Mainstream carries a vision to bring the latest happenings and insights to 8.2 billion people and to place technology at the centre of conversation for leaders navigating the future.