Generative AI music platforms operate fundamentally like samplers, recombining patterns from training data rather than creating entirely novel compositions. Initially, major AI music companies resisted licensing requirements by invoking 'fair use' doctrine, but lawsuits and market pressure are shifting behaviour, with an increasing number of firms beginning to secure proper licenses for their training datasets.

Drawing on decades of experience in music clearance, the author argues that established legal and licensing infrastructure developed around traditional music sampling can be effectively adapted to govern AI music generation. She notes that sampling itself was once dismissed as theft but eventually gained artistic legitimacy through standardised licensing protocols, and contends that AI proponents should similarly embrace licensing frameworks rather than pursuing fair-use arguments to achieve comparable professional legitimacy.