Copyright Protection of AI-Generated Music
By Anton Ylikallio[*]Jur. dr, Senior lecturer, University of Helsinki.
1. Introduction
AI-generated music has become a very dynamic field in recent years.[1]Notable publications on the copyright eligibility of AI-generated music include Axhamn, Johan “Copyright and Artificial Intelligence – with a focus on the area of music” in Morten Rosenmeier, Thomas Riis, Jens Schovsbo and Henrik Udsen (eds), Festskrift til Jørgen Blomqvist (Ex Tuto Publishing 2021) and Bulayenko, Oleksandr and Quintais, João Pedro and Gervais, Daniel J. and Poort, Joost, AI Music Outputs: Challenges to the Copyright Legal Framework, reCreating Europe Report, February 2022, available at https://ssrn.com/abstract=4072806. Not only have services like Suno and Udio achieved remarkable popularity,[2]Udio attracted 600,000 users within its first two weeks of release in April 2024 and was reportedly generating 864,000 songs per day within a month of becoming publicly available, see Ingham, Tim “The Train Has Left the Station: AI Music Platform Udio Is Already Spitting Out 10 Songs a Second”, 2024, available at https://www.musicb usinessworldwide.com/the-train-has-left-the-station-ai-music-platform-udio-is-already-spitting/. In November 2025, Suno was reported to generate 7 million songs daily, see Robinson, Kristin “Suno creates an entire Spotify catalog’s worth of music every two weeks says investor pitch deck for $250M fundraise”, 2025, available at https://www.billboard.com/pro/suno-creates-spotify-catalog-music-two-weeks-pitch-deck/, both links accessed 17.4.2026. but AI-generated compositions are increasingly making headway into the commercial market.[3]AI artists Xania Monet and Breaking Rust attracted notable attention and commercial success. See Eqbal, Amelia “AI Artist Xania Monet Has Hit the Billboard Charts. What Does It Mean for ‘Real’ Musicians?”, 2025, available at https://www.cbc.ca/news/entertainment/ai-artist-xania-monet-radio-billboard-chart-9.6967542; Simpson, Beth “The No. 1 country song in the US right now is AI-generated”, 2025, available at https://www.musicradar.com/music-tech/the-no-1-country-song-in-the-us-right-now-is-ai-generated. Both links accessed 17.4.2026. The pre-eminent example of the latter trend is the song Jag vet, du är inte min by artist Jacub, which topped Swedish charts on Spotify in January 2026 after garnering several million streams on the platform. It was later revealed that the song was AI-generated, which spurred debate on the relationship between human creators and AI-powered tools in the music industry.[4]See, inter alia, King, Ashley “Sweden Disqualifies Partly AI-Generated Folk-Pop Song After It Generates Millions of Streams Worldwide — Humans Behind the Song Defend AI as a ‘Tool In Our Creative Process’”, 2026, available at https://www.digitalmusicnews.com/2026/01/18/sweden-ai-folk-pop-song-banned/; Karlsten, Emanuel “Jacub svarar om AI-låten som toppar svenska Spotifys topplistor: Vi är ett kollektiv bakom”, 2026, available at https://emanuelkarlsten.se/jacub-svarar-om-ai-laten-som-toppar-svenska-spotifys-topplistor-vi-ar-ett-kollektiv-bakom/, both links accessed 16.4.2026. Regardless of how one views this discussion, AI-generated music is clearly set to grow in importance in the music industry, and the relationship between AI music outputs and copyright protection has arguably never been more relevant.
This article, which presents key findings of my doctoral dissertation,[5]Ylikallio, Anton “Musical Works, Copyright, and Generative AI”, Wolters Kluwer 2025, available at https://helda.helsinki.fi/items/7fd129f6-6a3e-4c83-afd6-be8b3d945cd6. examines whether and under what circumstances AI-generated musical compositions can receive copyright protection.[6]Some commentators make a distinction between “AI-generated” and “AI-assisted” creations, with the latter implying greater involvement by a human contributor in the creative process. I find this distinction difficult to uphold clearly and consistently, and I will therefore not use it. As aptly noted by Rognstad in “Copyright Protection for Output Resulting from Use of Generative AI Systems – Some Reflections”, NIR 1/2025, pp. 134–144, p. 144, a significant difficulty is that the use of generative AI “inevitably indicates that parts of the expression are generated by AI no matter whether AI is used as a tool of human creation or not”. The sole focus of the article is copyright protection of AI outputs, and it will not discuss other timely issues regarding AI and copyright law, such as training AI models on protected works or liability for AI-generated outputs.[7]Notable cases concerning these issues include Like Company v. Google Ireland (C-250/25); Kneschke v. LAION, OLG Hamburg, 5 U 104/24 (10.12.2025); GEMA v. OpenAI, LG München I, 42 O 14139/24 (11.11.2025); GEMA v. Suno, LG München I, 42 O 763/25 (2026). The article is divided into five chapters. After this brief introductory chapter, the second chapter presents currently available AI services and how they can be used to generate music. The third chapter examines the criteria for copyright protection and the main issues in applying these to AI-generated compositions. It notes that the distinction between protected expression and non-protected ideas is a particularly great challenge for protecting AI outputs, suggesting iterative prompting of an AI system as a potential means of securing protection. The fourth chapter examines the distinction between protected expression and non-protected ideas, compositional techniques and musical building blocks. It explores the use of quasi-statistical analysis and the experimental method of constructing “mock performances”, which synthesises expert musicological analysis with audio samples. The fifth chapter provides a summary and a concluding discussion.
