Songwriting is a revenue engine of the music industry. In 2023, Taylor Swift generated an estimated $75 million from streaming and radio airplay from her songwriting catalogue.[1]
Music publishers have structured the songwriting subindustry by organizing songwriting sessions, grouping songwriters on their rosters to pitch music to record labels and other artists, or to record if they also perform. For example, Rihanna’s “Diamonds” was written by Sia, Benny Blanco, Mikkel S. Eriksen, and Tor Erik Hermansen.[2] Though less recognizable, these writers create earworms people love. Sometimes, songwriting is an organic collaboration, where songwriters invite others to contribute artistically when they encounter writer’s block. For example, Cory Wong and Allen Stone’s “Look at Me” chorus was written by Cody Fry, who shared a demo that Wong and Stone expanded upon.[3]
From drum machines to sampling, technology is integral in the songwriting process. In the age of artificial intelligence (AI), the increasingly blurred intersection between technology and organic songwriting has shaken not only the music world but also its legal frameworks.
As musical compositions and sound recordings, the music on the market is copyrightable subject matter under the Copyright Act.[4] To be eligible for copyright protection, a work must be an original work of authorship.[5] In Zarya of the Dawn, the U.S. Copyright Office (USCO), where the office denied copyright registration for a comic book with AI-generated images, held that the selection and arrangement of images and accompanying text could contain sufficient original, human authorship to warrant copyright protection.[6] Following the USCO’s denial, Thaler v. Perlmutter opined that U.S. copyright law protects only works of human creation but noted when an AI tool is used to create a work, decisionmakers may ask “how much human input is necessary to qualify the user of an AI system as an ‘author’ if a generated work.”[7]
In January 2025, the USCO released a report stating “a work created solely through AI is not copyrightable, but a work that combines human creativity with AI can be copyrighted, so long as there is a ‘sufficient’ amount of human expression in that work.”[8] In October 2025, three major performing rights organizations (“PROs”) in North America, the American Society of Composers, Authors and Publishers (ASCAP), Broadcast Music, Inc. (BMI), and the Society of Composers, Authors and Music Publishers of Canada (SOCAN) announced that they will accept registrations of partially AI-generated musical works.[9]
This move bridges AI’s copyright implications and the integration of AI tools in music production software. For example, Logic Pro, a digital audio workstation (DAW), has integrated AI tools into its software like “session players”- on-demand, AI-generated instrumental parts that follow a track in the session. This function can enhance a songwriter’s efficiency, and the final product would qualify as partially human works for PRO registration..
However, the PROs and the Copyright Office have been silent on derivative works. The Copyright Act defines a derivative work as a work “based upon one or more preexisting works, such as a […] musical arrangement.”[10] The Act also elaborates that the copyright in a derivative work extends only to the material contributed by the author of such work.[11]
I recently encountered an AI-generated 1920s jazz version of Vybz Kartel and Spice’s Romping Shop, a 2009 dancehall hit. Though released as a parody, social media comments demanded its streaming release. Musical arrangements are usually afforded copyright protection, but when generative AI is involved, where will copyright’s protection and rights lie?
Since copyright in a derivative work extends only to the material contributed by the author of the derivative work, the character of the AI use’s character in creating a derivative work may may affect its copyright eligibility. One scenario is one the public encounters often, where an AI tool user feeds an entire work into an AI tool and has it create a musical arrangement from the user’s prompt, like the Romping Shop example. Another scenario is where a songwriter uses a fully AI-generated musical composition and/or sound recording as a basis for their new work by sampling it.
In the first scenario, one could argue that the copyright in the musical arrangement cannot extend to the material contributed by the author of the new work, because the author is not a human. However, if the user licenses the song from the original authors and commercially releases the musical arrangement, the USCO and PRO could block the arrangement’s registration lack of sufficient originality. Does that not infringe on the original author’s right to create a derivative work?[12]
In the second scenario, the ambiguous “partial” really burns. If the fully AI-generated sample were a bass riff, which a songwriter uses to create the basis of their work (like how Gnarls Barkley used Gianfranco and Gianpiero Reverbi’s Il carico d’oro bassline as the harmonic basis for the Crazy[13]), there may be a public domain-like analysis here. If the sample is ineligible for copyright protection, ownership would be limited to protectible elements of the final work. However, this kind of songwriting may create an idea-expression-like dichotomy – what if someone samples the new musical work and their derivative work contains that AI-generated sample? What rights would the songwriter who created the work first using the bass riff have in the newly sampled work?
The USCO and PROs should clarify the threshold of “partially AI-generated” works. While using AI tools to aid in creating an original musical composition aligns with the PROs’ purpose on its face, derivative works may present a different issue. Moreover, music publishers have an interesting dilemma: should they encourage their rosters of songwriters to refrain from using AI tools so that songwriting remains organically human? Or should they encourage their songwriters to use these AI tools to reduce the number of songwriters in a studio, which could increase their publisher’s share of the works’ public performance royalties?
[1] Glenn Peoples, Taylor Swift Grossed Almost $2b This Year From Her Music, Movie, Touring and Concert Merchandise, Billboard (Dec. 12, 2023), https://www.billboard.com/business/business-news/taylor-swift-earned-2-billion-music-movie-touring-1235555994/?utm_source=chatgpt.com.
[2] KiddFuture, Source TV: Benny Blanco Talks Producing Rihanna’s “Diamonds”, The Source (Oct. 7, 2012), https://thesource.com/2012/10/07/sourcetv-benny-blanco-talks-producing-rihannas-diamonds/.
[3] Cory Wong, ON THE ONE! // “Look At Me” (feat. Allen Stone), YouTube (Dec. 18, 2023), https://youtu.be/-YD-0TyvGLc?si=mZOO38biBgf5iEHV.
[4] 17 U.S.C. § 102 (a).
[5] Id.
[6] Copyright Review Board, Zarya of the Dawn (Registration # VAu001480196), United States Copyright Office (Feb. 21, 2023), https://www.copyright.gov/docs/zarya-of-the-dawn.pdf.
[7] Thaler v. Perlmutter, 687 F. Supp. 3d 140, 146, 149 (D.C. Cir. 2023).
[8] Daniel Tencer, Can AI-Generated Content Be Copyrighted? Here’s What a New Report From the US Copyright Office Says…, Music Business Worldwide (Jan. 30, 2025), https://www.musicbusinessworldwide.com/can-ai-generated-content-be-copyrighted-heres-what-a-new-report-from-the-us-copyright-office-says1/.
[9] Murray Stassen, ASCAP, BMI and SOCAN Will Now Accept Registrations of ‘Partially’ AI-Generated Musical Works, Music Business Worldwide (Oct. 28, 2025), https://www.musicbusinessworldwide.com/ascap-bmi-and-socan-will-now-accept-registrations-of-partially-ai-generated-musical-works/.
[10] 17 U.S.C. § 101.
[11] 17 U.S.C. § 103 (b).
[12] See 17 U.S.C. § 106 (2) (“the owner of copyright under this title has the exclusive rights to do and to authorize any of the following […] to prepare derivative works based upon the copyrighted work.”).
[13] Ben Williams, Influences: Gnarles Barkley, New York Magazine (Apr. 27, 2006), https://nymag.com/arts/popmusic/profiles/16845/.