Governing Artificial Intelligence: The Role of International Trade Law

Artificial intelligence (AI), which is projected to become the dominant technology in the decades to come, lies at the heart of the Fourth Industrial Revolution, the digital transformation that has had a profound impact on the development, production, consumption, and trade of goods and services in the XXI century. Inserting itself in the macro-debate on digital trade regulation and the treatment of emerging technologies under international economic law, this study focuses on a rather unexplored area in this field of law: the relationship between AI governance and international trade law. This research stems from the observation that, in response to the recent surge in AI applications, governments have started to adopt a patchwork of measures relevant to artificial intelligence that may affect the international supply and consumption of goods and services that rely on this technology. Acknowledging the key role that services play in the digital economy and recognising that AI may impact both services and services suppliers, this study focuses on AI- powered services to explore the role that international trade law can play in shaping AI governance. More specifically, it investigates the extent to which international trade agreements pose limits to the ability of governments to adopt AI-specific measures that can affect trade in services, and contributes to the establishment of AI governance frameworks. It does so by addressing four key issues: the nature and relevance of AI governance measures for trade in services; the impact of AI on the applicability of GATS disciplines; the potential limitations posed on AI governance by GATS obligations and commitments; and the relevance of new rules on digital trade in PTAs for AI governance. The research shows that AI is a peculiar digital technology that involves a strong link between cognitive capabilities and computer programs, and a combination of decades old theoretical concepts with new methods and practical applications. These features contribute to creating a certain degree of uncertainty regarding the applicability of multilateral rules on trade in services negotiated in the early 1990s. The study also reveals that governments’ approach to AI has changed remarkably since the founding of the WTO, with the period 1995-2015 characterised primarily by the adoption of policies that affect all digital technologies, including AI, and the period post-2015 defined by the introduction of policies that target AI explicitly and exclusively, such as AI ethical principles, AI standards, and AI-specific subsidies. While most scholarly work has already focused on the analysis of AI-related measures, this study focuses on AI-specific policies to assess the relationship between AI governance and international trade law. The study shows that, to the extent that a Member’s AI-specific policy is a measure affecting trade in services, it would fall under the purview of the GATS and be subject to its general obligations and specific commitments. However, a few factors may influence the scope and strength of the boundaries set by the GATS on AI governance, including the type of policy adopted, the entity implementing the policy, the rationale behind its design and application, the classification of AI-powered services, and the potential impact of AI on modes of supply and on the concept of likeness. The analysis also shows that digital trade rules emerging in PTAs can also contribute to shaping AI governance. Besides finding that international trade law can contribute to govern AI, this research also reveals that Al's disruptive nature may likewise influence international trade law moving forward, as it forces governments to reconsider the extent to which existing rules can keep up with technological progress, to evaluate the need for new disciplines that take into account the peculiar nature of Al, and to rethink the role of the WTO in the AI era.