The US is preparing to restrict the use of Chinese AI models by American companies. According to CNBC, a State Department representative said using such systems in corporate workflows raises “serious concerns”. Washington believes Chinese AI could promote Beijing’s position, censor unwanted topics and reflect the ideology of the Chinese Communist Party.
At the same time, businesses are looking more actively at Chinese models: they are cheaper and already look convincing enough for many work scenarios. CNBC cites Coinbase as an example: CEO Brian Armstrong mentioned using GLM 5.2 from Z.Ai and Kimi 2.7 from Moonshot. Earlier reports also said startup Lindy switched to DeepSeek to cut rapidly rising AI infrastructure costs.
Interest in Chinese models may also be growing because of instability in the US AI market. Nikkei Asia previously wrote that companies such as Airbnb and Uber accelerated adoption of Chinese solutions after Anthropic, at the request of authorities, suspended the use of two of its models — Mythos 5 and Claude Fable 5. For businesses, that was a signal: access even to major Western AI platforms may depend not only on contracts and pricing, but also on political decisions.
It is still unclear whether the US government can introduce a broad ban on private-sector model choice. Limiting its own government procurement is easier than forcing every company to drop Chinese AI, especially when open-source models are involved. In that scenario, legal disputes would almost certainly follow, including questions of free speech. Regulating foreign markets is even harder: Apple, for example, uses Alibaba’s generative AI platform for iPhones sold in China.
The AI conflict between the US and China looks set to intensify. According to WSJ, Chinese authorities are also discussing limits with local companies on using domestic models abroad. In addition, a Chinese vulnerability database linked to the relevant regulator claimed that Anthropic’s Claude Code allegedly contains backdoor-type vulnerabilities and poses a “serious threat”.