China Weighs Keeping Its Most Advanced AI Models at Home

Chinese authorities have held meetings over the past month with the country's leading model developers to discuss restricting foreign access to its most capable AI systems, according to reporting relayed by AI Post and AI ML Big Data, and independently detailed by Time. Alibaba, ByteDance, and the startup Z.ai were among those invited.

According to a summary published in an official Supreme People's Court journal, participants proposed a tiered framework. Basic open-source tools would need only a simple filing, more advanced technologies would face security reviews, and the most sensitive frontier models would be barred from public release or restricted to domestic use. Officials also discussed making the leak or theft of proprietary model weights an offense under China's national security law.

The move would reverse the approach that made Chinese labs globally competitive. Alibaba's Qwen and ByteDance's Doubao are among the most downloaded open-weight models worldwide, and Z.ai's GLM-5.2 has drawn attention for approaching US frontier quality at a fraction of the cost. Curbing their export would place model weights in the same controlled-export category that Washington already applies to advanced chips.

What is verified is that the discussions occurred and that a tiered proposal exists in an official journal. What is not settled is the scope or the timing. The reporting stresses that the rules may apply only to future models, and it is not clear whether or when they would take effect. Both governments now benefit from a framing in which their own technological frontier is worth protecting.