China’s Stance on AI and the Global Implications

China has made its position on artificial intelligence quite evident. Reports indicate that Beijing is looking to restrict foreign access to its advanced AI models. This move comes as second-tier Chinese companies rapidly adopt these cutting-edge AI technologies to attract global clients who may be affected by U.S. export restrictions. In China’s view, frontier AI is considered a national asset that requires protection, rather than just another marketable product.

There’s nothing surprising about this. Authoritarian regimes typically aim to control information and technology, and the associated profits. The U.S. shouldn’t fall into the same trap.

Interestingly, American technological superiority doesn’t stem solely from secrecy. It results from creating desirable products, establishing trustworthy standards, and outpacing competitors in growth.

As Congress discusses the National Defense Authorization Act, there are calls for fresh AI export restrictions. This is understandable; advanced technology can indeed empower foreign militaries and cyber attackers, which raises legitimate concerns.

However, Congress needs to differentiate between curtailing adversaries’ dangerous capabilities and maintaining a trustworthy global AI ecosystem with the U.S. at the forefront. This distinction should inform any export policy on AI.

The first principle focuses on national security through technological leadership. America’s AI firms, cloud service providers, chip manufacturers, educational institutions, and investment markets drive its global leadership. Export controls must safeguard this system rather than cripple it.

Secondly, allies and trustworthy partners should have special access to AI resources. The goal shouldn’t be to build an AI fortress but rather an AI coalition. Countries like Japan, South Korea, Australia, Taiwan, and New Zealand should not face stringent restrictions comparable to those imposed on China, Russia, Iran, or North Korea, provided they meet security and reversal criteria.

Trustworthy nations won’t base their developments around American technology if they fear that the U.S. can arbitrarily restrict access. They will seek alternatives.

Thirdly, controls need to be based on risk and be technically precise. The U.S. is already employing focused measures to keep risky technology out of the hands of unsuitable actors. These measures prove most effective when tailored to specific end users or particular applications.

If companies aren’t clear on what the law allows, they will struggle to perform effectively. This is exacerbated by the remote access security legislation currently under Senate consideration.

There are indeed loopholes regarding remote access. If companies can’t obtain advanced chips but can access similar computational capabilities through the cloud, then current export controls fail.

The legislation aims to close this loophole. It would enable the Department of Commerce to regulate remote access for foreign nationals via networks or cloud services if it poses a notable risk to U.S. national security.

This could provide regulators with greater discretion.

The term “foreign nationals” could extend to individuals and businesses from allied nations, while “foreign policy” could be interpreted flexibly. Together, these factors might yield broad powers for the executive branch over global access to American AI capabilities.

This bill would require the Department of Commerce to inform Congress about prospective regulations, including security concerns and potential economic impacts. However, Congress wouldn’t need to approve these regulations before implementation.

This arrangement might simplify administration, but it may not sufficiently reshape the global AI landscape.

The rapidly changing tech industry demands some degree of regulatory discretion, though not unlimited authority. The ambiguity within the bill is likely to influence corporate behavior long before any court challenges its scope.

Companies often become overly cautious when unsure of who might get access. Cloud providers could end up excluding broad segments of foreign clients, AI institutes might shy away from international partnerships, and universities could stall on research collaborations. Consequently, reliable allies and legitimate enterprises could be treated like adversaries.

If customers in the U.S. can’t access AI, they will turn to other options—many of which now lead to China. Companies like DeepSeek, Alibaba, Huawei, Tencent, and Kuaishou are actively courting developers and investors around the globe. Each unnecessary hurdle Congress imposes on legitimate AI users in the U.S. presents an opportunity for these Chinese firms.

The economic stakes are substantial. Estimates suggest AI could inject up to $15.7 trillion into the global economy by 2030, with China anticipated to be the primary beneficiary. Meanwhile, U.S. private AI investments are projected to surpass $285 billion by 2025, significantly outpacing China’s overall investment.

This lead is remarkable but not guaranteed to last. Customers are naturally inclined toward ecosystems that are capable, accessible, dependable, and politically stable. America’s competitive advantage could diminish if U.S. regulations create uncertainty while China makes access to second-rate products easier.

Congress should not abandon the Remote Access Security Act but rather work to enhance it. A more effective bill would eliminate loopholes while ensuring safe pathways for legitimate usage. Key priorities should include granting allies privileged access, clarifying thresholds for frontier-scale computations, and differentiating between various uses of AI technology.

Legislators should establish timelines for licensing, review processes, safe harbors for companies engaging in due diligence, and regular assessments to adapt controls as technology evolves.

This strategy doesn’t imply leniency towards China; instead, it enables the U.S. to confront China without undermining its own position.

The U.S. must deny adversaries the tools necessary for military advancements while also simplifying secure transactions and complicating those that pose risks.

China aims to control the AI race, but a smarter approach for the U.S. involves disciplined openness—making it easier for allies to leverage American AI while defending against potential threats.

Congress should craft export regulations that safeguard U.S. interests instead of unintentionally conceding them to Beijing.

This approach has kept American companies at the forefront of global technology for decades, and it’s how they can continue to do so.