A report circulating this week says China is extracting outputs from American artificial intelligence systems and using them to help train its own defense technology.
The claim comes from Roya News, surfaced via Google News under the headline "China extracting US AI outputs to train defense systems." The available source material is limited to that headline, so the specifics — which US systems, which Chinese programs, how the material was obtained, and on what evidence — are not detailed in what has been published so far. Readers should treat the scope of the allegation as unconfirmed until fuller reporting appears.
What is on the record is Beijing's stated ambition. According to the Economic Times, President Xi Jinping — who also heads the military — told a Politburo group study session on Thursday that the People's Liberation Army should "strengthen the military application of unmanned intelligent technologies." The outlet framed the remarks as a push to build what it called an "intelligent military," with Xi urging the PLA to deepen development in the area.
Put side by side, the two items sketch a familiar pattern in the US-China technology contest: an explicit, top-level Chinese commitment to militarized AI, paired with an allegation that American AI systems are an input into that effort.
The technical premise is worth understanding. Modern AI models can be trained partly on text and other outputs generated by existing models — a practice sometimes described as distillation. That makes a capable commercial AI system potentially useful to a rival even without access to its underlying code or weights, which complicates the export controls and chip restrictions Washington has leaned on.
It matters because if a country's AI lead can be partly copied through the answers its models give away, controlling hardware alone may not control the technology.