China is considering a crackdown that could turn AI technology leaks into national security crimes. According to people familiar with the matter, regulators from the Ministry of Commerce and the National Development and Reform Commission (NDRC) have held multiple closed-door meetings over the past month with Alibaba, ByteDance, and Zhipu AI. Officials discussed restricting overseas access to China's most advanced AI models to prevent core technology from leaking abroad.
The meetings also covered a range of preventive measures. Officials proposed that any leak or theft of proprietary large-model technology could be classified as a crime endangering national security and subject to the strict national security law. To cut off technology outflows through financial channels, the government is also considering limiting foreign investment in domestic AI startups.
A compliance crackdown on startups moving overseas has already begun. In April, regulators ordered Meta to withdraw its $2 billion acquisition of Manus, a China-backed AI startup, and launched compliance investigations into Manus and other startups attempting to move abroad, checking whether they violated export control laws. In early June, the government issued new regulations further tightening overseas deals involving Chinese investors, technology, data, and national security.
Earlier, legal experts had proposed a three-tier classification for open-source large models: basic open-source tools would only need to be filed, advanced models would require security reviews, and the most sensitive frontier models would be restricted to domestic use.