The Artistic Vision: A Fusion of Past and Future
Just dropped a reimagined version of The Magic of You, a tune I wrote and released with The Ink Spots over 40 years ago. This redux features Kalina, an AI-generated artist born from Mark Seidenfeld’s Papua New Guinea “Mud Queen” photography. Her face and voice were synthesized using Suno (audio) and Pixnova (visuals). The result? A surreal blend of nostalgia and cutting-edge tech, now live on my YouTube channel at https://youtube.com/jammmerpro.
This project sits where creativity, technology, and law converge. Kalina isn’t human, but her creation is grounded in human authorship: the “Mud Queen” image as visual source, my original song as musical foundation, and intentional curation of AI tools under our creative direction. Lyrics are unchanged. Melody is reimagined. Voice is animated, never imitated.
Legally Speaking: This Is a Derivative Work
This is a derivative work, full stop. Section 101 of the U.S. Copyright Act defines a derivative work as one “based upon” preexisting works, including those “recast, transformed, or adapted” [17 U.S.C. § 101]. That’s exactly what we’ve done.
I own the sound recording (SR) and performance (PA) rights to the original The Magic of You. Mark Seidenfeld owns the copyright in the “Mud Queen” photo. We fed these controlled works into AI systems to produce new expression. This isn’t incidental, it’s foundational. Every element of Kalina originates in or responds to our inputs.
The U.S. Copyright Office’s 2025 Report on Copyright and Artificial Intelligence, Part 2: Copyrightability confirms this: when human-authored works are used as expressive inputs to AI, and appear in the output, the result builds upon the original[3]. The Office notes that creators using AI to “refine, recast, or modify” their preexisting works may have legitimate authorship claims[3]. In our case, it’s not “some circumstances”, it’s the entire premise.
AI Styling Is Not Copyrightable, But Integration Matters
AI-applied “style” alone isn’t copyrightable. Style, genre, or algorithmic texture, without copying protected expression, falls outside protection. Copyright doesn’t protect ideas, processes, or methods [17 U.S.C. § 102(b)]. AI stylistic application is a process.
Yet, AI modification of melody or vocal performance doesn’t negate derivation. AI is a tool, like a synthesizer or auto-tune. As the Copyright Office states, “the use of AI tools to assist rather than stand in for human creativity does not affect copyright protection for the output”[3]. My creative choices (i.e. setting parameters, selecting outputs, integrating voice and image) constitute human authorship for a derivative work.
De Minimis and the “Alphabet of Music” Revisited
I asked: “If you cut a single note with protected timbre, is it de minimis, or just the alphabet of electronic music?” This alludes to the 2024 Second Circuit ruling in Structured Asset Sales, LLC v. Sheeran, No. 23-905[1].
There, the court affirmed that the chord progression and harmonic rhythm in “Let’s Get It On” were too commonplace to be protectable under the 1909 Act. It excluded elements not in the deposited sheet music and rejected the plaintiff’s “selection-and-arrangement” theory as a matter of law.
The de minimis doctrine varies by circuit: the Sixth Circuit holds that any unlicensed sample, even a two-second chord, can infringe (Bridgeport Music, Inc. v. Dimension Films)[2], while the Ninth Circuit requires “practical importance or economic value” (VMG Salsoul v. Ciccone)[2].
But in our case, we’re not sampling others, we’re transforming our own work. So de minimis doesn’t defend; it defines. If only one note were borrowed, it might be trivial. Here, the entire framework is derived from owned assets. The AI didn’t invent Kalina, it derived her. Nothing trivial about that.
AI Authorship: The Human Hand Remains Central
Crucially: purely AI-generated material is not copyrightable. Copyright requires human authorship. But when a human creatively controls (i.e. selecting, modifying, arranging, and integrating AI output) the work can be protected, especially as a derivative work[3].
In this case:
• I directed the musical transformation.
• Mark authorized use of his “Mud Queen” image.
• We co-curated Kalina’s emergence, choosing prompts, rejecting outputs, refining results.
This is authorship by design. As the Copyright Office states, “the putative author must be able to constrain or channel the program’s processing of the source material”[3]. That’s what we did.
Implications of Input: Everything Derived Is, Well, Derived
WHEN YOU FEED PREEXISTING CONTROLLED CONTENT TO AI, EVERYTHING PRODUCED IS DERIVED FROM THE ORIGINAL. This is legally sound when the input is protected and substantively present in the output.
The Copyright Office recognizes that feeding copyrighted material into AI triggers derivative creation questions[3]. But here, where I own the music and Mark owns the photo, we’re not infringing. We’re licensing our own works internally to spawn a new version.
This doesn’t license others to do the same. Unauthorized use of my original recording to create a competing AI version would be infringement. Here, the circle is closed: rightsholders are also creators.
Final Thoughts: Evolution, Not Revolution
As both producer and attorney, I’m not torn. I’m energized. This isn’t legal uncertainty. It’s creative clarity. The Copyright Office confirms existing doctrine accommodates this innovation[3]. No new law needed.
This redux isn’t nostalgia wrapped in AI, it’s evolution. A dialogue between decades. A testament to music, imaging, and intellectual honesty about authorship.
Would love your take. And yes, I want you to listen. 😉
BTW, my friends still quote The Three Faces of Eve when I dive into music, tech, and law. Don’t worry. We’re all still fine.
Best wishes,
Yoda, JammerPro, and EZ
References:
[1]: https://law.justia.com/cases/federal/appellate-courts/ca2/23-905/23-905-2024-11-01.html
[2]: https://www.talksonlaw.com/blog/a-split-on-de-minimis-copyright-infringement-in-music
[3]: https://www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-2-Copyrightability-Report.pdf
[4]: https://www.supremecourt.gov/DocketPDF/24/24-981/351412/20250311112410889_Structured%20Asset%20Sales%20LLC%20Corrected%20Petition.pdf
[5]: https://www.loeb.com/en/insights/publications/2024/11/structured-asset-sales-llc-v-sheeran
[6]: https://caselaw.findlaw.com/court/us-2nd-circuit/116663909.html
[7]: https://www.supremecourt.gov/search.aspx?filename=/docket/docketfiles/html/public/24-981.html
[8]: https://www.musicbusinessworldwide.com/files/2024/11/SAS-vs-Sheeran-1.pdf
[9]: https://www.govinfo.gov/app/details/USCOURTS-ca2-23-00905
[10]: https://natlawreview.com/article/lets-not-get-it-battle-greatest-hits
[11]: https://www.courthousenews.com/wp-content/uploads/2024/04/SAS-sheeran-appellee-brief.pdf
[12]: https://www.youtube.com/watch?v=18ZhfA6uUzE