China slams US claims of ‘industrial-scale’ AI theft

The response comes after three US agencies accused Chinese AI companies of exploiting American AI models.

A humanoid robot on display at the Chinese International Supply Chain Expo (CISCE) in Beijing [File: Kevin Frayer/Getty Images]

China has issued a forceful reply against claims by the United States that top Chinese artificial intelligence companies have engaged in “industrial scale” theft of their US counterparts’ AI models.

The response on Wednesday comes after three US federal agencies – the Federal Bureau of Investigation (FBI), National Security Agency (NSA) and the Cybersecurity and Infrastructure Security Agency (CISA) – accused Chinese AI companies of exploiting American AI models to benefit their own systems.

list 1 of 3 OpenAI unveils latest AI model amid rising scrutiny and safety concerns

list 2 of 3 As the China-US AI race reaches Cairo, Egypt faces a strategic decision

list 3 of 3 AI researcher quits Anthropic saying AI race ‘could kill us all’

The joint advisory on Tuesday said Chinese companies like DeepSeek, Alibaba and Moonshot AI had distilled “billions of tokens” from US models since “at least late 2024”, likely with the Chinese government’s “awareness”.

Tokens are the smallest units of text that AI models read, or generate.

On Wednesday, China’s Commerce Ministry dismissed the allegations and accused the US of seeking a “monopoly of the AI industry”.

“If the U.S. suppresses Chinese AI companies under the pretext of targeting distillation, China will take resolute countermeasures,” the ministry said in a statement to the Associated Press.

Distillation is a method in which lesser AI models are trained on more advanced ones. The ministry said the practice is commonplace in the AI industry, including by US AI companies. But American firms say the volume in which it’s being done by Chinese AI models amounts to stealing.

“The sheer scale of these campaigns and their sophistication indicate that distillation is not a supplement to these companies’ AI model development, but the critical core of it,” the US advisory claimed.

The AP reported that China’s Ministry of Foreign Affairs called on the US to “refrain from making unfounded accusations or smears” against China.

“China’s AI development is the result of high-level technological self-reliance and strength,” ministry spokesperson Mao Ning told reporters at a news conference. “We maintain that all parties should strengthen cooperation to promote AI development that is open, inclusive, universally beneficial and oriented toward the common good.”

The US advisory comes just weeks before Chinese President Xi Jinping is set to visit the US, where President Donald Trump will host a summit on a range of topics, including AI.

In a rare move, Xi could be accompanied on his US visit by a large number Chinese executives, according to the Reuters news agency. It is not clear which executives would join the delegation, but past meetings between the two countries may indicate AI leaders could be in the mix.

Xi brought Alibaba founder Jack Ma during his 2015 visit to the US. When Trump visited Xi during a trip to Beijing in May, the US president was joined by more than a dozen American executives, including Tesla CEO Elon Musk.

Under Trump, the US government has taken a hardline against efforts to advance Chinese AI models. China is aggressively investing heavily in artificial intelligence and robotics, pouring nearly $300bn into what the government sees as the country’s next economic engine.

While speaking in Dallas on Tuesday, US Treasury Secretary Scott Bessent told an audience that China can “never get ahead” of the US in AI.

“The Chinese distill our models and they can never get ahead of us,” Bessent said.

The Chinese companies named in the US advisory did not immediately respond to Al Jazeera’s request for comment.

此前，美国三个机构指责中国人工智能公司利用美国人工智能模型牟利。

在北京举行的中国国际供应链博览会（CISCE）上展出的人形机器人 [图片：Kevin Frayer/Getty Images]

美国声称中国顶尖人工智能公司“大规模”窃取美国同行的AI模型，中国对此予以了强有力的回应。

周三的回应是在美国三个联邦机构——联邦调查局（FBI）、国家安全局（NSA）和网络安全与基础设施安全局（CISA）——指责中国人工智能公司利用美国人工智能模型来使其自身系统受益之后做出的。

列表 1/3：OpenAI 在日益严格的审查和安全担忧中发布最新 AI 模型

（共3条，第2条）随着中美人工智能竞赛蔓延至开罗，埃及面临一项战略抉择

3/3 人工智能研究员辞职 Anthropic 公司，称人工智能竞赛“可能会毁灭我们所有人”

周二发布的联合咨询报告称，自“至少从 2024 年底开始”，DeepSeek、阿里巴巴和 Moonshot AI 等中国公司已经从美国模式中提炼出“数十亿代币”，这很可能是中国政府“知情”的。

词元是人工智能模型读取或生成的最小文本单元。

周三，中国商务部驳斥了这些指控，并指责美国试图“垄断人工智能产业”。

“如果美国以打击蒸馏技术为借口打压中国人工智能公司，中国将采取坚决的反制措施，”该部在发给美联社的一份声明中表示。

蒸馏是一种利用更高级的人工智能模型来训练较低级别人工智能模型的方法。该部表示，这种做法在人工智能行业很常见，包括美国的人工智能公司也在使用。但美国公司认为，中国人工智能模型大量使用蒸馏技术无异于窃取成果。

美国咨询机构声称：“这些攻击活动规模之大、手段之高表明，数据提炼并非这些公司人工智能模型开发的补充，而是其关键核心。”

美联社报道称，中国外交部呼吁美国“停止对中国进行毫无根据的指责和抹黑”。

“中国的人工智能发展是高度技术自主和实力的成果，”科技部发言人毛宁在新闻发布会上对记者说。“我们坚持认为，各方应加强合作，推动人工智能发展成为开放、包容、普惠、面向共同利益的事业。”

美国发布此警告之际，正值中国国家主席习近平即将访问美国前几周，美国总统唐纳德·特朗普将在美国主持召开峰会，讨论包括人工智能在内的一系列议题。

据路透社报道，习近平此次访美行程或将罕见地携众多中国企业高管随行。目前尚不清楚哪些高管会加入代表团，但两国以往的会晤记录表明，人工智能领域的领军人物或许也在考虑之列。

习近平在2015年访美期间带上了阿里巴巴创始人马云。今年5月，特朗普访问北京期间会见了习近平，陪同他的是十几位美国企业高管，其中包括特斯拉首席执行官埃隆·马斯克。

在特朗普执政期间，美国政府对推进中国人工智能模型研发采取了强硬立场。中国则大力投资人工智能和机器人技术，向政府视为国家下一个经济引擎的领域投入了近3000亿美元。

美国财政部长斯科特·贝森特周二在达拉斯向听众表示，中国在人工智能领域“永远不可能领先于”美国。

“中国人会提炼我们的模型，他们永远无法超越我们，”贝森特说。

美国咨询文件中提到的中国公司尚未立即回应半岛电视台的置评请求。