China plans to tighten export controls on open-weight models and 23 key AI technologies
decoder ·
China's commerce and science ministries have released a draft regulation to tighten export controls on 23 key AI technologies, including frontier models, AI agents, and autonomous malware tools. Large language models face restrictions if they were trained on sensitive Chinese data or meet specific thresholds such as having at least 2 trillion parameters, utilizing 10^26 FLOPs of training compute, or ranking in the global top 10. While these measures aim to restrict foreign access to critical capabilities, the plan exempts technology that is already publicly available through open-source releases or research papers. To support its domestic open-source ecosystem, China introduces a fast-track licensing process that grants automatic approval for registered open-weight releases. The regulations distinguish between cloud-based API usage, which remains unrestricted, and the installation of restricted tech on local servers abroad, which requires standard export licenses. This shift may compel US companies to rely more heavily on domestic models while creating significant challenges for European efforts to maintain AI sovereignty using open-weight systems. Beijing also intends to facilitate AI development in the Global South through the World Artificial Intelligence Cooperation Organization.
decoder