Author
Assoc Prof Ezieddin Elmahjub
Organisation/Institution
Qatar University
Country
QATAR
Panel
Information Technology Law
Title
A Comparative Legal Analysis of Global Governance Models for AI Training Data
Abstract
The use of copyrighted works to train generative AI models has become the central fault line in contemporary copyright and technology law. Around the world, legislatures, courts, and regulators are experimenting with different ways to reconcile large-scale text and data mining with the rights and expectations of authors. Yet we lack a systematic framework for understanding how these emerging solutions fit together and what they mean for innovation, justice, and cross-border coordination. This paper develops a comparative, data-informed typology of governance models for AI training data across a selection of influential jurisdictions: the United States, European Union, United Kingdom, Japan, Canada, China, Australia, Singapore, and India. Using a qualitative comparative legal analysis (QCLA) methodology, the study systematically collects and codes post-2024 statutes, regulations, leading judicial decisions, and policy reports dealing with AI training, copyright exceptions (such as fair use, fair dealing, and text and data mining exceptions), transparency obligations, and opt-out mechanisms. From this coded corpus, the paper inductively identifies and refines three dominant governance models: a judicial adjudication model grounded in open-textured doctrines (for example, fair use), a regulated opt-out model that combines statutory TDM exceptions with enforceable opt-out rights for creators, and a purpose-based exception model that turns on the functional use of data (such as non-enjoyment or non-consumptive use). Each model is evaluated along several dimensions: legal certainty versus flexibility, default rule (opt-in versus opt-out), enforcement burdens, and transactional efficiency. Asia is integrated into the analysis in two ways. First, several Asian jurisdictions (Japan, China, Singapore, India) are treated as full case studies within the typology. Second, the paper’s normative discussion examines how global fragmentation in AI training data rules shapes Asia’s options for building sustainable innovation ecosystems, ensuring informational justice for creators, and pursuing coherent regional integration in digital trade and data governance.
Biography
Dr. Ezieddin Elmahjub is an Associate Professor of Law at Qatar University and a Senior Research Fellow at Harvard Law School for the 2025–2026 academic year. He previously served as Chair of the Centre for Law and Development (CLD), where he led major interdisciplinary initiatives at the intersection of law, ethics, and technology. He holds an LLM and a PhD from Queensland University of Technology, Australia. His academic career includes appointments at Swinburne University, Queensland University of Technology, the University of New England, and the National University of Singapore, reflecting broad experience across common law and mixed legal systems. Dr. Elmahjub’s research focuses on AI governance, data ethics, and the capacity of Islamic jurisprudence to address contemporary legal and technological challenges. He is the author of An Islamic Vision of Intellectual Property (Cambridge University Press, 2019), a landmark contribution to comparative legal theory. His work appears in leading journals such as Philosophy & Technology, the Oxford Journal of Law and Religion, the Asian Journal of Comparative Law, and IEEE Technology & Society. He has led pioneering projects on the ethical design of AI systems, the use of predictive modelling in judicial institutions, and the regulation of extended-reality environments in the metaverse. His current research explores global governance models for AI training data, the normative implications of generative AI, and the integration of Islamic legal reasoning into modern regulatory frameworks. Dr. Elmahjub is widely recognised for bridging doctrinal analysis with technological insight, and for advancing rigorous, culturally grounded approaches to the governance of emerging technologies.
