China Is Now Blocking Top AI Engineers at Alibaba and DeepSeek from Leaving the Country

China is now restricting overseas travel for top AI professionals at private firms such as Alibaba and DeepSeek — a significant expansion of controls that previously applied mainly to state-owned enterprises, university researchers, and nuclear scientists . The measure, reported by Bloomberg on May 26, 2026, signals an escalation in Beijing's campaign to prevent technology outflows and close the AI gap with the United States .

Who is affected and how the controls work

Government agencies have begun imposing travel restrictions on individuals involved in advanced AI work who are assessed as strategically important to the country. The affected categories include startup founders, senior researchers and engineers, and corporate executives . Under the new regime, these individuals must secure approval from relevant authorities before leaving the country .

In earlier, more targeted cases, the enforcement mechanisms were even more direct. After DeepSeek’s R1 model captured global attention in January 2025, several of the company’s employees had their passports confiscated by its parent entity, High-Flyer, with government support . Zhejiang provincial authorities also began screening investors before they could meet with DeepSeek leadership and instructed headhunters to stop recruiting from the firm .

The criteria remain deliberately opaque

Reports indicate that inclusion on a travel-restriction list hinges on whether an individual is "working on advanced AI" and considered to have "strategic importance" to the nation . Chinese-language coverage has emphasized that this is not a simple rank- or title-based filter. Instead, it depends on a broader government assessment of national value — a judgment opaque enough to cover a wide swath of the private AI workforce .

State-backed narratives have begun framing these engineers as "national treasure"-level assets whose knowledge-transfer could hand foreign competitors — particularly U.S. firms — an unfair advantage . The implication is clear: Beijing now views leading AI talent the way it has long viewed nuclear physicists or senior state-enterprise executives.

What the travel bans reveal about Beijing’s strategy

The expansion of exit controls is not an isolated policy. It forms one leg of a wider regulatory apparatus designed to tighten control over technology outflows.

Preventing talent-led leakage at the source. Earlier in 2025, Beijing issued verbal instructions to restrict the export of key technologies and the movement of skilled workers . The travel bans harden those verbal instructions into enforceable administrative barriers that operate directly on individuals.

Policing the boundary between domestic innovation and foreign capital. The Manus case set an important precedent. When Meta agreed to acquire the agentic-AI startup Manus for roughly $2 billion in late 2025, Chinese authorities imposed exit bans on its co-founders and launched a formal review under technology-export and outbound-investment rules . The review, announced by the Ministry of Commerce in January 2026, was designed to determine whether the deal complied with China’s technology-import and -export regulations . The episode demonstrated that Beijing will use exit controls not merely to retain people, but to block attempts to "Singapore-wash" assets or relocate intellectual property abroad .

Securing technological sovereignty through human capital. By treating AI engineers as strategic state assets, Beijing signals that it considers talent the hardest-to-replace component of its AI competitiveness. The Carnegie Endowment noted that DeepSeek’s breakthrough renewed party confidence but also triggered intensified oversight — a pattern now extending across the private sector . A separate analysis described this as embedding mobility controls across the entire administrative apparatus and into the private sector, with judicial and administrative exit bans now applicable to private-sector founders and ordinary citizens .

Sustaining momentum in the US–China AI race. The timing is telling. Chinese open-weight models — led by Alibaba’s Qwen family — surpassed U.S. developers in Hugging Face downloads between August 2024 and August 2025, capturing 17.1% of all downloads compared to 15.8% for U.S. developers . By locking in its top talent, Beijing aims to sustain that momentum while denying the U.S. a pipeline of Chinese-trained AI experts who could accelerate American research. Early evidence suggests the policy may already be forcing individual career choices, pressuring engineers with international ambitions to decide early whether to stay in China or leave before they become restricted .

The limits of what is known

Despite the breadth of the reports, significant gaps remain. It is unclear exactly how many employees are affected, which specific job levels or roles trigger a restriction, or whether the policy will be applied uniformly across the industry . This ambiguity itself appears to serve a strategic purpose, maximizing the chilling effect while maintaining administrative flexibility.