The global artificial intelligence landscape is rapidly fracturing, as both the United States and China increasingly view AI access through the lens of national security and geopolitical competition. A recent analysis by the New York Times DealBook highlights how the two superpowers are actively building technological walls around their most advanced AI systems, a development that threatens to dismantle the collaborative open-source philosophy that has driven much of the industry’s recent progress. As these walls rise, third countries and international developers find themselves caught in the middle of an escalating tech war.
The trend toward restriction is already visible. According to the NYT, some officials in the Trump administration are actively pushing to limit foreign access to Chinese open-weight models, reinvigorating efforts to add Chinese AI labs to the Commerce Department’s Entity List — a move that would effectively cut off US access without a license. The National Security Agency and the White House Office of the National Cyber Director have also reportedly considered issuing an advisory practically discouraging US companies from using Chinese AI tools.
Critics of open-weight Chinese models, including Anthropic’s Dario Amodei and OpenAI’s Sam Altman, have pushed for mandatory government testing and even licensing requirements for frontier models. Prominent voices on the other side, including venture capitalist Bill Gurley and former White House AI czar David Sacks, have pushed back, arguing that open-source AI represents legitimate competition rather than a security threat.
China Contemplates Its Own Export Controls
The US actions have not gone unanswered. As reported by the Financial Times, Beijing is now actively considering imposing its own export controls on advanced AI models and domestically produced chips. The Chinese Ministry of Commerce has reportedly consulted with major tech firms, including Alibaba and ByteDance, about restricting overseas access to their models. This represents a significant pivot for China, whose tech sector has largely relied on an open-weight strategy to gain global market share and challenge American dominance.
The impetus for these potential Chinese restrictions appears to be the realization that highly capable open-weight models, such as DeepSeek and Moonshot AI’s Kimi K3, are being utilized by foreign entities, including US military and intelligence agencies. By contemplating limits on the transfer of training data and the downloading of model weights by foreign users, Beijing is signaling that it, too, views frontier AI as a strategic asset that must be protected from geopolitical rivals.
The End of the Open-Source Era?
The mutual weaponization of AI access marks a profound shift in the trajectory of technological development. For years, the rapid advancement of AI was fueled by a culture of open research and cross-border collaboration. However, as AI systems become increasingly powerful and their potential military and economic applications become clearer, both Washington and Beijing are prioritizing security over openness.
This fragmentation poses significant challenges for the global tech ecosystem. Developers in Europe, Asia, and elsewhere who have relied on open-source models from both the US and China may soon find their access curtailed. The bifurcation of the AI landscape into separate, walled-off spheres could slow the overall pace of innovation, increase costs, and complicate efforts to establish global standards for AI safety and governance. As the NYT analysis suggests, the walls being built today will shape the technological balance of power for decades to come.
The Global South Caught in the Middle
For developing nations in the Global South, the mutual weaponization of AI access presents a particularly acute dilemma. Many of these countries have built their nascent AI strategies around open-weight models from both the US and China, viewing them as a cost-effective path to technological leapfrogging. If both superpowers restrict access to their most capable systems, these nations may find themselves locked out of the frontier AI era entirely, widening the global technological divide.
China’s recent initiative at WAIC, the formation of a global AI governance body with 29 founding nations, predominantly from the Global South, can be read in this context as a strategic effort to position Beijing as the more open and collaborative partner. Washington has intensified its push to expel Chinese open-weight AI from corporate America, making Beijing’s open-access posture all the more strategically calculated. By offering training programs, governance frameworks, and access to Chinese AI tools, Beijing is cultivating goodwill and influence in regions that Washington has largely neglected.
The NYT analysis notes that this competition for global AI influence may ultimately prove as consequential as the race for raw model performance, as the country that sets the standards and shapes the norms of AI governance will exercise enormous soft power over the technology’s trajectory for decades to come.
The irony of the current moment is that both the US and China are simultaneously arguing for openness in AI governance while restricting access to their most powerful systems. Washington champions democratic values and open markets while tightening export controls; Beijing promotes multilateral AI governance while consulting on domestic restrictions. For the rest of the world, navigating this contradiction will require careful diplomacy and a clear-eyed assessment of which partnerships best serve their long-term technological and economic interests.
The walls being built today are not yet permanent, but the longer they stand, the harder they will be to dismantle. The NYT analysis concludes with a sobering observation: in a world where AI capabilities are increasingly tied to national security, the default assumption of openness that has characterized the technology industry for decades is giving way to a new paradigm of strategic restriction, one that will require new international frameworks and norms to manage effectively.