Cross-Border Model Training and Data Sovereignty: Conflicts-of-Law Across the AI Supply Chain

Abstract

Artificial Intelligence (AI) model training increasingly depends on vast, globally distributed datasets, creating unprecedented challenges for cross-border governance. Data sovereignty laws such as the EU’s General Data Protection Regulation (GDPR), China’s Data Security Law, and emerging U.S. state privacy statutes—directly impact how AI developers collect, process, and transfer training data across jurisdictions. This article examines conflicts-of-law issues that emerge in the AI supply chain, particularly regarding data localization, algorithmic accountability, intellectual property, and trade secrecy. By engaging with case law, treaties, and academic scholarship, it highlights the tension between innovation and sovereignty in transnational AI model training. Drawing on comparative examples from the United States, European Union, and Asia, the paper explores theoretical frameworks of extraterritoriality, conflict-of-laws doctrines, and corporate liability in the AI ecosystem. It also considers ethical implications, including fairness, transparency, and the impact of surveillance-driven AI. The article concludes by recommending policy reforms—such as harmonized treaties, stronger international cooperation, and accountability mechanisms—to balance sovereignty with innovation.