{"pk":63024,"title":"Play It Again, HAL: Evaluating Fair Use in Generative Music Artificial Intelligence Training","subtitle":null,"abstract":"<p>This paper evaluates fair use in the context of the training process for generative music AI systems, such as those creating text-to-music, voice-to-music, instrumental-only, lyrics-only, and other outputs. Training data for such systems is comprised of musical compositions and sound recordings, much of which is under copyright. This paper considers the four fair use factors and how courts may weigh them in favor of, or against, fair use in the unauthorized copying of copyrighted works for music AI training. This paper adopts the approach outlined in <em>Andy Warhol Found. for the Visual Arts v. Goldsmith</em>, 598 U.S. 508 (2023) for the first fair use factor, which emphasizes proper framing of the specific use and the purpose of the allegedly infringing secondary work at issue—here, the generative music AI system and its use by end-users. This paper will consider three possible views of the purpose of a generative music AI system (as a tool, as entertainment, and as functional music) and illustrate how each framing may influence the analysis of each fair use factor and the ultimate result of the fair use inquiry.</p>","language":"eng","license":{"name":"All rights reserved","short_name":"Copyright","text":"© the author(s). All rights reserved.","url":"https://creativecommons.org/licenses/authors"},"keywords":[],"section":"Article","is_remote":true,"remote_url":"https://escholarship.org/uc/item/1bs5j3fg","frozenauthors":[{"first_name":"Susan","middle_name":"","last_name":"Wang","name_suffix":"","institution":"","department":""}],"date_submitted":null,"date_accepted":null,"date_published":"2026-02-18T03:58:00+01:00","render_galley":null,"galleys":[{"label":"PDF","type":"pdf","path":"https://journalpub.escholarship.org/uclalaw_elr/article/63024/galley/48678/download/"}]}