Conference Agenda

The use of artificial intelligence (A.I.) in the creation of music and other forms of art has become a hot-button topic in recent years. There is significant concern, both within the musical community and among the general public, that unregulated A.I.-generated music threatens the livelihoods of music creators, both through improper use of copyrighted music as training data for A.I. models and through the competition created by the outputs of such models. Yet, to date, many countries, including the U.S., have no legislation regulating A.I.-generated music, nor has any U.S. court ruled on the copyright implications of the musical training data or output of an A.I. model.

While this absence of legal guidance may feel counterintuitive given the rapidly developing capabilities of A.I. technology, it is a familiar phenomenon in the history of music copyright law: changes in the law have often failed to keep up with changes in technology. In the period of legal uncertainty following technological innovations, musical practices often incorporate and build on the technological innovation such that, by the time the law catches up, its regulations disrupt established practices of making music. Copyright law’s prolonged battle with sampling is one of the most recent examples of this pattern. But is copyright law’s historic inability to adapt to new technologies of music-making doomed to repeat itself now with A.I., or can it learn from the mistakes of the past?

This paper provides a broad overview of the landscape of copyright law relating to A.I.-generated music. It begins with a discussion of proposed legislation in the United States aimed at regulating the use of A.I. in creative works as of November 2025. It then addresses the implications of the Copyright Office's guidelines on the copyrightability of works created using A.I. It concludes with discussion of the parallels between the rise of A.I.-generated music and the historic rise of sampling with the goal of avoiding the belated, practice-disruptive court rulings that now govern sampling.

In 2021, a new generative AI startup called Soundful launched, offering users a simple and streamlined process for generating royalty-free background music for their social media posts. Soundful provides their users with the right to sell, sub-license, distribute or stream the products of their inputs as stand-alone music tracks, dependent on the licensing agreements that they sign. In this paper, I argue that Soundful’s licensing model circumvents both the human as artist and the human as subscriber as the result of a fundamental discrepancy between Soundful’s commodification of intellectual property rights and the US Copyright Office’s current policies, which presently limit legal protections available to products of generative AI. In other words, music’s commodity status is supplanted, from Soundful’s point of view, and the copyright itself becomes a tradable commodity. Thus, music becomes the intermediary through which copyright is commodified.

By bringing the US Copyright Office’s novel and developing perspectives into conversation with Soundful’s licensing model, I highlight an irreconcilable disagreement between how intellectual property rights are commodified by Soundful and how legal protections under copyright law are misappropriated for the products of generative AI. As public policy develops to address AI and IP (often lagging far behind technological innovation), it is necessary to promote a human-centered regulation of training models. Focusing on Soundful as a case study, I evaluate the ethical implications of copyright as a commodity in the generative AI space. By evaluating my findings in parallel to current intellectual property policy developments in the United States, I will foreground a proactive approach to the assessment of and resistance to potential harms of generative AI.

Digital technologies have offered opportunities and challenges for the creation and distribution of music. The music industry has had a difficult transition to digital era economic, business, and cultural realities. These recent challenges bear similarities to past eras, including in the early twentieth century, when the advent of records, player pianos, and other new technologies posed a significant challenge to existing artistic, legal, institutional, and business assumptions and arrangements.

Uses of artificial intelligence in music highlight continuing technological challenges for music. Technological challenges for music also underscore core sources of continuing tension for the application of copyright to music. Copyright, which initially protected literary works, reflects a visual-textual bias that has long contributed to the inexact fit of copyright for music. Pattern and repetition in music are core sources of this inexact fit. Many musical genres are characterized to a significant degree by particular and at times even predictable configurations of notes that may reflect pattern in chord sequences, melodies, and polyphony, for example. Pattern and repetition in music, the centrality of performance to music, and the relational construction of music contribute to difficulties in determinations of infringement in music copyright cases. These difficulties are often evident in music copyright infringement analysis, including in doctrines such as substantial similarity, that seek to determine how much copying is too much copying. Determinations about copyright in music cases may be at considerable tension with forms of musical expression that may be repetitive and pattern-based.

Notably, artificial intelligence, which typically refers to a collection of technologies that seek to make machines more human-like, is also often pattern-based. Pattern recognition is a core element of machine learning technologies and algorithms that enable systems to identify patterns, make decisions, and improve themselves. Machine learning is just one branch of artificial intelligence that incorporates pattern recognition. The centrality of patterns and pattern recognition in artificial intelligence systems and music may suggest potential paths for conceptualizing and approaching music copyright infringement cases.