The New Songwriters: AI, Music and the Law

In 2023, “Heart on My Sleeve,” a song with Drake and The Weeknd’s voices, began racking up views on Spotify. The problem was that neither Drake nor The Weeknd released or performed the song. It was artificially generated based on hours of their music.

An industry turned on its head

While the music world has been familiar with AI for some time, generative AI’s ability to compose new music has turned the industry on its head. These tools are used not only to enhance existing music, but also to create “original” music with just a few prompts. Boomy, MusicFlow, Soundraw, Soundful and Suno are just a few of the many AI-powered music creation tools available online enabling users to create new music based on existing copyrighted music. This advancement raises concerns about copyright ownership, credit to musicians whose music is used for AI training and consent from those musicians. On some of these platforms, users can create songs in particular genres, about certain subject matter and even in the style of an artist — using the artist’s voice and work without their consent.

Copyright status of AI-generated music

Musical works, their sound recordings and lyrics qualify for copyright protection. The fundamental requirement of copyright law is that copyright protection can only be acquired by works created by human authors. However, in February 2025, the Copyright Office granted copyright protection for the first time to an artwork generated by AI because it was made using sufficiently creative human prompts. Invoke, a generative AI platform and author of the work, kept note of every prompt to argue that there was human authorship in the work. They received copyright protection only in the selection, coordination and arrangement of the piece by the human author. By this logic, AI-generated music can be copyrighted.

Attribution, compensation and licensing arrangements

One central issue with AI-generated music is attribution, because prompts such as “create a song in the style of Van Morrison” or “create a country song about a brown eyed girl” imply taking both small and large portions from other artists’ lyrics, expression and sounds—sometimes without the prompter’s knowledge or intent. Attribution is more complicated because the AI is trained on the work of countless artists, making it difficult to trace or credit individual contributors. Some companies purport to resolve this attribution issue. For example, Soundraw’s website claims there is no risk of harming artists, as the company trains its AI on music that Soundraw created itself. Lack of attribution dovetails with lack of compensation — copyright holders now must find a way, through licenses and perhaps legislation, to ensure they can be compensated for their work.

AI-generated music tools all have different rules regarding intellectual property ownership. For example, OpenAI claims intellectual property rights, including copyright, in its proprietary tools, software, models, algorithms and systems and then licenses the work to the prompter. Open AI does not claim ownership of the output. However, the license agreements may also differ, which means users should investigate the terms of use and licensing of the program they wish to use prior to creating music.

Liability of music generators for copyright infringement

Whether AI can infringe is still an unsettled area of law. As recently as June 2025, the Northern District of California ruled that it was fair use for Anthropic (a tech company) to use subsets of purchased versions of copyrighted books to train the large language model (LLM) that powers its AI service, in large part because the use was transformative. The plaintiffs in that case negotiated a $1.5 billion settlement in September 2025 because Anthropic permanently stored some pirated works, although they arguably may be entitled to more than $1 trillion in damages. The settlement is still awaiting approval and could change.

AI tools are susceptible to infringement because many are trained on LLMs that require massive amounts of data. Furthermore, many of the AI music generator companies have indemnity clauses on their websites that release them from liability if their users are sued for songs they create using the companies’ platforms. One of these websites, MusicFlow, has a notice on its website requiring users to notify them of any copyright infringement, which may optimistically signal that the platform is taking steps toward accountability and responsible use of content. If companies were trained only on songs in the public domain or previously licensed works, the risk of copyright infringement would be low.

UMG Recordings, Capital Records, Sony Music Entertainment and other music labels filed a complaint against AI music generator Suno in June 2024, alleging that “the foundation of [Suno’s] business has been to exploit copyrighted sound recordings without permission” and that Suno has willfully infringed “on an almost unimaginable scale.” (Complaint, pg. 3) The music labels claim “that using targeted prompts that include the characteristics of popular sound recordings — such as the decade the sound recording was released, as well as the topic, genre, and descriptions of the artist — can cause Suno’s product to generate music files that strongly resemble the Copyrighted Recordings related to the descriptions in the prompt.” (Complaint, pg. 16) Suno responded, claiming fair use. This case and other similar suits remain unresolved. For instance, in 2023 Concord Music Group sued Anthropic for $75 million, alleging infringement of musical works in its catalog. For a master list of pending AI and copyright cases, see ChatGPT Is Eating the World.

Liability of music generators under right of publicity and other laws

In addition to copyright infringement claims, there may be other legal recourse against AI music generators. One such claim is the right of publicity doctrine, which prevents unauthorized use of someone’s name, image or likeness for commercial purposes, which is broadly interpreted to mean anything that evokes an individual, including their voice.

In cases where AI convincingly recreates a deceased singer’s voice, another legal protection may apply. Several states, including New York, California, Nevada, Texas, Florida and Hawaii, recognize a postmortem right of publicity through specific statutes. However, in New York, California and Texas, the postmortem right of publicity applies only to individuals deemed to have ‘commercial value.’

Conclusion

AI-generated music devalues the human voice and capitalizes on artists’ work without compensation. Others note that AI-generated music is a tool for artists to enhance their music and create music faster, more cost-effectively and in greater quantities than before. For instance, the CEO of SoundExchange likened AI-generated music to the advent of music videos in that it creates a new way for musicians to showcase their creativity.

Whatever your situation, Lutzker & Lutzker can help you protect and manage your music and publicity rights in the age of AI. For further reading from our website on the topics discussed here, see the following Insights and IP Bits & Pieces®: Music Video Copyright, Compilations v. Individual Copyright Registrations of Photographs, AI Can’t Hold Copyrights and our Copyright FAQs.