China has begun restricting overseas travel for top artificial intelligence professionals working at private companies. The measures now reach firms such as Alibaba Group Holding Ltd. and DeepSeek. Government agencies require selected individuals to obtain official approval before leaving the country.
People familiar with the matter described the expansion as an escalation. It aims to protect sensitive technology while accelerating efforts to close the gap with the United States. The curbs apply to startup founders, researchers and executives whose work Beijing deems critical to national interests. What stands out is the extension into the private sector. Authorities once focused such limits on state-linked entities.
But the policy has roots. The Next Web reported that controls first surfaced at DeepSeek earlier this year. Staff there surrendered passports after the lab’s R1 model challenged assumptions about compute requirements for frontier AI. Engineers and researchers received requests to hand over documents. The stated reason involved potential access to state or commercial secrets.
Similar warnings circulated. Since early 2025 Chinese authorities have advised leading AI experts against travel to the U.S. National security concerns and risks of information leakage drove those messages. ThinkChina detailed how the Manus case amplified tensions. The China-founded AI firm relocated operations to Singapore in 2024. When Meta considered a major acquisition Chinese regulators investigated. The company’s CEO and chief scientist received exit bans.
That episode sent ripples. Other startups faced regulatory warnings. Investors erected internal firewalls to limit cross-border movement of staff and data. And the pattern continues. Bloomberg News revealed on May 26 2026 that the government now evaluates individuals based on their specific contributions rather than title or employer alone. Approval processes add friction. Affected professionals must clear trips with relevant authorities. Short. Direct. No appeal mentioned.
The shift coincides with financial tightening. In late April agencies instructed leading AI developers including Moonshot AI StepFun and ByteDance to reject U.S.-origin capital without prior clearance. Several firms now weigh reincorporating from offshore jurisdictions back into mainland China. Domestic initial public offerings appear more favorable under the new signals. Capital talent and corporate structures all pulled inward.
Numbers tell part of the story. Stanford’s 2026 AI Index showed the performance gap between top U.S. and Chinese models narrowed to 2.7 percent. That compares with 17.5 to 31.6 percentage points in mid-2023. China accounts for 69.7 percent of global AI patents. It produces 23.2 percent of AI publications worldwide. Industrial robot installations run nine times higher than in the United States. Talent flows tell another tale. Migration of AI specialists to the U.S. has fallen 89 percent since 2017.
Yet the talent pool carries complexities. More than five million AI professionals represent a reported shortage according to estimates from China’s Ministry of Human Resources and Social Security cited in People’s Daily in 2024. Roughly half of companies surveyed by Beijing News say they lack multidisciplinary expertise. Returnees have increased. From 12 percent of overseas-trained Chinese AI researchers returning after NeurIPS 2019 the figure rose to 28 percent by 2025 per analysis in The Economist.
High-profile moves illustrate the trend. Pan Zizheng a computer science PhD chose Hangzhou-based DeepSeek over Nvidia in 2023. The model later topped free app charts in the U.S. store. Nvidia scientist Yu Zhiding reposted praise noting many top talents originate from China. Harvard professor Graham Allison posed the question why such advances happened domestically rather than in American labs. Other returnees include Wu Yonghui former Google DeepMind vice president now at ByteDance Vinces Yao ex-OpenAI now at Tencent and Zhou Hao former head of reinforcement learning on Google’s Gemini team who joined Alibaba.
Moonshot AI founder Yang Zhilin followed a familiar path. Tsinghua undergraduate Carnegie Mellon PhD stints at Google Brain and Meta then return in 2023. These professionals often cite competitive salaries reaching two to three million RMB for senior technical roles. National narratives also resonate. Zhu Song-Chun who spent 28 years in the U.S. returned in 2020 to found the Beijing Institute for General Artificial Intelligence. He has spoken of historical duty referencing Qian Xuesen’s legacy. “I could never forgive myself” if the country needed him he once said.
But frictions exist. Intense competition and “involution” wear on some. Commercialization trails Silicon Valley. University-industry pathways feel less fluid. Wu Yi returned from OpenAI to Tsinghua in 2020 as part of the so-called Berkeley Four. By April 2025 he departed for Meta again. The university removed his profile. Handover disputes and teaching feedback turned negative.
Broader forces shape decisions. U.S. export controls visa restrictions and scrutiny of Chinese-linked researchers have pushed some talent home. China’s maturing large language models and university research funding pull others. Beijing treats frontier AI as strategic. The passport requirements function in practice as informal exit bans. They carry no judicial review. International collaboration which once strengthened Chinese academic work now faces complications.
Enforcement grows harder as the list expands. What began with a handful at DeepSeek now touches researchers across private labs. Thousands potentially. Firms stay silent. Neither DeepSeek nor Moonshot AI commented publicly on the latest reports by late May 2026. The policy’s effectiveness remains untested at scale.
Analysts watch the U.S. side too. American immigration policies have tightened H-1B approvals. Security reviews alienate some Chinese scientists. Brookings Institution analysis from 2024 warned these steps threaten U.S. leadership by restricting talent inflows. Chinese-origin researchers still form a large share of top talent. Roughly 38 percent of leading AI experts in the U.S. completed undergraduate studies at Chinese universities according to Paulson Institute data.
So Beijing doubles down. Recent Bloomberg reporting underscores the intent. Safeguard what has been built. Accelerate what remains unfinished. The AI race has moved beyond chips and data centers. It now targets the minds that design the systems. Travel approvals represent one lever. Funding rules another. Corporate domicile a third.
Private innovation drove much of China’s post-ChatGPT progress. DeepSeek Alibaba Moonshot and peers produced models that narrowed gaps quickly. Yet that same dynamism now triggers controls. The government sees leakage risks in conferences board meetings or even casual conversations abroad. Executives could face detention as leverage. Intellectual property might transfer through acquisitions or partnerships.
Results appear mixed so far. Performance metrics improve. Talent returns accelerate. Yet creativity demands openness. Some researchers chafe at restrictions. Others accept them as necessary in a geopolitical contest. The coming years will test whether such measures consolidate advantage or stifle the very progress they seek to protect.
One fact stands clear. The world’s two largest economies have turned AI talent into a domain of strategic competition. Passports sit with employers. Approvals sit with agencies. And the flow of ideas faces new gates.
China Locks Down AI Stars at Private Labs in Bid to Outpace U.S. Tech Rivals first appeared on Web and IT News.