MAIN CONTENT
Disclosure instead of prohibition
The proposal described by Reuters would ask businesses to post information about their use of Chinese AI on their own websites. It would not, on the information currently available, impose a general ban on those models. That distinction matters: disclosure changes the information visible to customers, investors and policymakers, while leaving companies formally able to choose among tools. Whether such a policy would apply to every internal experiment, only customer-facing systems or particular high-risk applications has not been established.
Reuters said discussions involved the Treasury Department and the White House Office of Science and Technology Policy, but the administration has not publicly set out a legal mechanism, enforcement standard or timetable. A debated idea is not an enacted obligation. Companies cannot reliably assess its compliance burden until officials clarify what counts as Chinese AI, which aspects of a model's origin matter, and how frequently disclosures would have to be updated.
Why companies turn to Chinese models
Chinese-developed models such as Qwen, Kimi and DeepSeek are attractive partly because they can deliver useful performance at a comparatively low cost. Open-weight releases also allow developers to download and adapt model parameters, although the availability of weights does not automatically disclose the training data, original source code or all security properties.

















