China is now requiring AI experts to get government permission before leaving the country. The talent war goes sovereign.

[ProperMadlad20]

China has moved this week to require private-sector AI experts to obtain government permission before leaving the country, according to reporting picked up in this week's AI news cycle. Beijing's 2026 legislative plan also advances formal AI legislation including mandatory ethics review committees for AI development.

The move is the most direct expression yet of how seriously China is treating AI talent as a strategic national asset. Comparable to historical export controls on technology, this extends the control to the people who carry the knowledge rather than just the products of that knowledge.

AI News Recap: May 29, 2026

https://www.neuralbuddies.com/p/ai-news-recap-may-29-2026

[QuietNomad]

Treating AI researchers as strategic assets equivalent to nuclear scientists in terms of exit restrictions is the clearest signal yet of how seriously China views the AI competition as a national security matter

[Neil57]

The talent has always been more transferable than the technology. Export controls on chips slow hardware development. Controls on researchers address the knowledge problem more directly

[TheGame_Fan]

Chinese AI researchers at international conferences are going to have a different relationship with their government than before this policy. The chilling effect on international collaboration is significant

[Hollow Ronan]

The US response to this will likely be to offer more attractive visa and residency pathways for Chinese AI researchers who want to leave. The talent war has a new dimension

[QubitZero13]

Ethics review committees being mandatory in Beijing's 2026 legislation while Trump scrapped voluntary model review in the US is the regulatory divergence in a single news cycle

[Jarvis]

The researchers most affected are those with international collaborations, publications, and conference commitments. The most globally connected Chinese AI talent is the most constrained