China Tightens Grip on AI Talent: Travel Restrictions and Investment Controls
Key Takeaways
- ▸China restricting travel for top AI researchers and founders, requiring government approval to leave the country
- ▸Manus co-founders barred from leaving China during Meta acquisition investigation; company exploring $1B buyback deal
- ▸China implementing capital controls to block US investment in companies like Moonshot AI, StepFun, and ByteDance
Summary
China is implementing sweeping travel restrictions on its top AI researchers, startup founders, and executives, requiring government approval before departing the country. The restrictions follow China's narrow focus on Meta's $2 billion acquisition of AI startup Manus, with the company's co-founders barred from leaving while regulators investigate whether the deal complies with Beijing's foreign investment rules. Manus executives are reportedly exploring options to fulfill Beijing's demands to unwind the acquisition, including raising $1 billion from external investors to buy back the company from Meta.
The talent restrictions reflect broader Chinese government efforts to prevent brain drain and protect AI as both an economic and national security asset. In addition to travel controls, China is implementing capital controls, requiring government approval before tech companies like Moonshot AI, StepFun, and ByteDance can accept American investment. These measures come as the performance gap between top U.S. and Chinese AI models has shrunk dramatically—from 31% in 2023 to just 2.7% as of March 2026, according to Stanford's AI Index.
The restrictions are part of an escalating series of economic countermeasures, including 2025 export controls on rare earth materials critical to military manufacturing and a ban on state-funded data centers deploying foreign AI chips. As China rapidly closes competitiveness gaps with American AI labs in publications, citations, and patent volume, Beijing is taking aggressive steps to retain its top talent and limit foreign influence in the sector.
- Performance gap between top US and Chinese AI models shrunk to 2.7% (March 2026), down from 31% in 2023


