Rebalancing Rights: how Generative AI forces a rethink of fair use/ fair dealing under Canadian and USA law

The rapid emergence of generative artificial intelligence has unsettled longstanding assumptions in copyright law, challenging both the United States’ flexible fair use doctrine and Canada’s more structured fair dealing framework. As AI systems rely on large-scale ingestion and reproduction of copyrighted material for model training, they expose doctrinal gaps in how each jurisdiction conceptualizes reproduction, transformation, substantiality, and user rights. This article examines how generative AI disrupts traditional copyright norms by blurring the boundaries between analytical computation and derivative expression. Through a comparative analysis of U.S. and Canadian law, it highlights the contrasting flexibility of fair use and the purpose-based constraints of fair dealing, showing how both regimes struggle to accommodate large-scale machine learning processes. The article argues that neither framework designed for human-centered creativity adequately addresses the economic, moral, and control interests of creators in the AI era. It proposes a recalibration of copyright exceptions and recommends the adoption of a statutory AI Training Exception incorporating transparency, auditing, and compensation mechanisms. Such reform, it contends, is essential to rebalancing the rights of creators, users, and innovators while preserving technological progress and legal certainty in the age of generative AI.