The artificial intelligence race between the United States and China has entered a new phase, and it is not about who has the better model or the faster chips. It is about who gets to keep their people.
According to a Bloomberg News report published this week, Chinese authorities have quietly expanded travel restrictions to cover top artificial intelligence professionals working at private companies. The policy, which applies to individuals working on what the government classifies as "strategically important AI projects," requires researchers to obtain official permission before leaving the country. The companies affected include some of China's most prominent technology firms, among them Alibaba and DeepSeek.
For the global technology industry, this is a significant escalation. What began as warnings and advisories has crystallized into a formal gatekeeping mechanism designed to prevent the movement of human capital. In the context of an industry where talent is the single most valuable resource, the implications are far-reaching.
From Advice to Enforcement
The current restrictions are not without precedent. In March 2025, Beijing reportedly "advised" senior executives and researchers at leading AI companies against traveling to the United States. The concern at the time was threefold: the risk of data leaks during international conferences and meetings, the threat of technology theft by foreign intelligence services, and the persistent danger of talent poaching by American competitors offering salaries that Chinese firms struggle to match.
Those advisories, however, were informal. What has changed is the shift from recommendation to requirement. The new policy reportedly requires researchers to submit travel requests to government authorities and wait for approval before booking flights or applying for visas. The criteria for approval remain opaque, but sources familiar with the matter suggest that travel to the United States and its allied nations faces the highest scrutiny, while trips to Belt and Road partner countries may proceed with fewer obstacles.
The mechanism places Chinese AI researchers in a position analogous to defense contractors or nuclear scientists in Western countries, whose travel is monitored and controlled due to national security concerns. The message from Beijing is unmistakable: artificial intelligence is no longer merely a commercial sector. It is a strategic asset, and the people who build it are strategic assets too.
The Companies in the Crosshairs
Alibaba and DeepSeek were specifically named in the Bloomberg report as companies whose personnel are affected by the new requirements. Both firms occupy critical positions in China's AI ecosystem, albeit in different ways.
Alibaba, one of China's technology giants, has invested heavily in large language models through its Qwen series and has integrated AI capabilities across its cloud computing, e-commerce, and logistics operations. The company's research teams work on everything from multimodal models to agentic systems, and its cloud platform serves as infrastructure for thousands of Chinese AI startups.
DeepSeek, meanwhile, has emerged as one of the most closely watched AI laboratories in the world. The Hangzhou-based firm burst into global consciousness in early 2025 when its R1 reasoning model demonstrated capabilities that rivaled top American systems at a fraction of the training cost. DeepSeek's subsequent models have continued to attract attention for their efficiency and performance, and the company has become a symbol of China's ability to innovate in AI despite American export controls on advanced semiconductors.
The travel restrictions place both companies in a difficult position. International collaboration has been a cornerstone of AI research since the field's inception. Major breakthroughs have historically emerged from cross-pollination between laboratories in different countries. Conferences like NeurIPS, ICML, and ICLR serve as essential venues for sharing research, recruiting talent, and building the personal networks that drive scientific progress. If Chinese researchers cannot attend, or must navigate a bureaucratic approval process that may take weeks or months, their ability to participate in the global scientific conversation is fundamentally compromised.
Part of a Broader Pattern
The travel restrictions fit into a wider framework of Chinese industrial policy aimed at achieving technological self-sufficiency while protecting domestic advantages. In the semiconductor sector, where American export controls have limited China's access to advanced chipmaking equipment, Beijing has poured billions into domestic alternatives. The results have begun to show: according to IDC, Chinese chipmakers now control 41 percent of the domestic AI accelerator market, a dramatic increase from just a few years ago when Nvidia dominated the landscape.
The government has also taken a more interventionist approach to corporate transactions involving foreign buyers. Earlier this year, Chinese regulators blocked Meta's proposed acquisition of Manus AI, an agentic artificial intelligence startup, reportedly over concerns that the deal would transfer strategically valuable technology to an American company. The move surprised industry observers, who had expected the transaction to proceed with standard antitrust review.
Together, these measures paint a picture of a country that views artificial intelligence not as a commercial sector subject to normal market forces, but as a domain of national competition where the state has both the right and the obligation to intervene. The United States has adopted a similar posture through its export controls, investment screening mechanisms, and restrictions on Chinese access to American cloud computing resources. Both countries are treating AI as a zero-sum competition, and human talent is increasingly viewed as a resource to be hoarded.
The Talent Dilemma
For individual researchers, the new restrictions create immediate professional complications. Many Chinese AI scientists hold dual affiliations, working part-time at domestic companies while maintaining visiting positions at foreign universities or collaborating with international colleagues on joint projects. The travel restrictions could make such arrangements untenable, effectively forcing researchers to choose between their Chinese employment and their international connections.
The policy also raises questions about recruitment. Chinese AI companies have historically been able to attract top talent from overseas by offering competitive salaries, large research budgets, and the chance to work on problems of national significance. If those positions come with the implicit cost of restricted mobility, the calculus changes. A researcher who knows they may be denied permission to attend a conference in San Francisco, present a paper in Montreal, or visit collaborators in London may think twice before signing a contract in Beijing or Hangzhou.
Conversely, the restrictions may strengthen retention among researchers who are already in China. The so-called "reverse brain drain" — in which Chinese scientists who studied and worked abroad return home — has been a notable trend in recent years. Travel restrictions make it harder to leave, but they also signal that the government considers the work valuable enough to protect. For some researchers, that recognition may offset the inconvenience of restricted mobility.
Global Reactions
The international response to China's tightening controls has been muted but pointed. American technology executives, speaking privately, have expressed concern that the restrictions will deepen the bifurcation of the global AI ecosystem into competing spheres of influence. In this scenario, research published in China would increasingly be conducted by researchers who do not attend international conferences, do not collaborate with foreign colleagues, and do not participate in the informal networks that currently lubricate scientific exchange.
Some analysts see a longer-term risk to China's competitiveness. The history of science suggests that openness and exchange are powerful drivers of innovation. The Soviet Union maintained formidable scientific capabilities in areas like mathematics and physics, but its isolation from Western research communities meant that its scientists often solved problems that had already been solved elsewhere, or pursued avenues that had been abandoned for good reason. If China's AI researchers are cut off from the global conversation, they may face similar inefficiencies, no matter how talented they are individually.
Others take a different view. They point out that DeepSeek's breakthroughs emerged precisely in an environment of restricted access to American technology, suggesting that constraints can breed creativity rather than stifle it. The argument holds that Chinese researchers, forced to find efficient solutions rather than brute-forcing problems with the latest Nvidia chips, have developed techniques that are both innovative and practically valuable.
What Comes Next
The travel restrictions are likely to be just one of several measures China deploys to control its AI talent pool. Rumors have circulated for months about tighter regulations on foreign ownership of Chinese AI startups, expanded requirements for domestic data storage, and new rules limiting the ability of Chinese citizens to work remotely for foreign companies. None of these have been formally announced, but the direction of travel is clear.
For the global AI community, the challenge is adapting to a world where the free movement of researchers — once taken for granted — is increasingly constrained by geopolitical competition. International conferences may need to create hybrid participation models that allow Chinese researchers to present remotely. Companies may need to design collaboration frameworks that do not require physical co-location. Academic journals may need to reconsider their peer review processes to account for reviewers who cannot meet authors in person.
The deeper question is whether artificial intelligence can continue to develop as a genuinely global field under these conditions. The optimistic view holds that digital collaboration tools, open-source software, and preprint servers can maintain scientific exchange even when human movement is restricted. The pessimistic view warns that AI research depends on the kind of informal, trust-based relationships that form best in person, and that restricting travel will inevitably slow the pace of discovery.
Neither Beijing nor Washington appears likely to relax its grip. Both countries have concluded that artificial intelligence will shape the balance of power for decades to come, and both are willing to sacrifice some degree of scientific openness in pursuit of strategic advantage. The researchers caught in the middle — the people whose work could, in a different world, be shared freely for the benefit of humanity — are increasingly being treated as national resources rather than global citizens.
For now, the AI professionals at Alibaba, DeepSeek, and China's other leading technology firms are waiting for permission to board their flights. Whether those permissions are granted, and on what terms, will be one of the defining variables in the next chapter of the global artificial intelligence race.