Following the United States, is China also considering restrictions on AI models?...Will the 'sovereign AI' trend speed up as AI border walls become reality?

Following the US trend toward tighter controls, Chinese authorities are reportedly reviewing measures that could restrict overseas access to advanced AI models built in the country. Reuters, citing local sources, said on 7 July (local time) that the Chinese government discussed potential restrictions with Alibaba, ByteDance, and Z.ai (ZhipuAI). Any move would likely start with future next-generation models. The article notes that many Chinese developers use an “open-weight” approach, allowing models to be downloaded and used more freely, a strategy interpreted as a way to expand market share. After the US issued export-control directives on Anthropic’s advanced models last month, China’s view appears to have hardened, treating AI models as a national strategic asset similar to semiconductors. Performance comparisons cited include Z.ai’s GLM-5.2 scoring 51 versus Google’s Gemini 3.5 Flash at 50. Some analysts expect a “dual structure” to keep general models open while controlling cutting-edge releases, which could affect cost-conscious buyers.