From Scarcity to Scale: Cross-Border Data Flows for Artificial Intelligence

This paper examines how cross-border data flows—defined as the transmission of data across national boundaries—can serve as a catalyst for scientific progress in artificial intelligence (AI). The analysis begins by establishing the critical role of both the quantity and quality of data in developing advanced AI systems. As global AI models approach the upper limits of available human-generated data, cross-border exchanges become increasingly vital to expand and diversify training datasets. The paper demonstrates that such flows can improve data representativeness, helping mitigate geographic and demographic biases in AI development. It then turns to governance, identifying the regulatory approaches adopted by countries to either facilitate or constrain international data transfers. These approaches range from unrestricted flows to full data localization, often driven by considerations around national security and personal data protection. The paper offers a comparative analysis of the regulatory regimes in China, the European Union, and the United States—three major jurisdictions with large influence over global data governance. Despite diverging on certain restrictions, these blocs share overlapping instruments such as standard contractual clauses and certification mechanisms, which can serve as potential avenues for regulatory convergence. Building on these overlaps, the paper offers three recommendations to foster alignment: (1) leveraging existing common instruments, (2) creating a global repository of governance frameworks to identify shared practices, and (3) embedding standardized data-sharing mechanisms into multilateral or trade-based agreements. The paper concludes that cross-border data flows, if effectively governed, can increase global access to diverse, high-quality datasets, ultimately accelerating the scientific advancement of AI.