Three Los Angeles songwriters say they wrote it first.
Steve Cooper, Jon Sandler, and Greylyn Johnson sued HYBE, Big Hit Music, Artist Publishing Group, and writers of BTS hit “Swim,” explaining that they finished a demo by the same name in March 2025, and circulated it to publishers through the platform Disco.ac. A year later, “Swim” came out on BTS’s album “Arirang.” The members of BTS are not defendants, BTW. Those are the facts as I understand them. After that it gets murky.
Here’s BTS “Swim.”
Right up front, there’s one considerable hurdle for Musicologize: hearing the plaintiffs’ demo. It was never released, and I haven’t found it. The filing in the public record, however, tells me a lot. I can speculate, and I will, but there are many unknowns. “Swim” is out there, and I’m familiar with it, and the complaint contains musical transcriptions (sheet music) of the plaintiffs’ work along with, presumably, the strongest musical arguments the plaintiffs can make that point to copying. I’m absolutely inclined to accept the notes as presented in those transcriptions. We’ll start there and consider how well they support the conclusions put forth in the complaint.
Quick refresher on copyright infringement and musicology
Copyright infringement, I’ve written here many times, is a bit of a three-legged stool. The three legs are access, similarity, and originality. A plaintiff needs all three. Access is considered first because infringement requires copying — you can’t copy what you’ve never had access to. That would be “coincidence,” and in the world of copyright, coincidence is effectively the opposite of “infringement.” You then need to find similarity in the two works, enough to matter, and those similarities need to be sufficiently original and protectable by the copyright owner of the earlier work. If the similarities are commonplace and found in lots of other works, the third leg fails.
The access case
Its access leg appears sturdy, timestamped receipts of who at the publisher played the file and when. Disco.ac analytics attached as Exhibit A allegedly show Artist Publishing Group people opening and playing the file. The complaint says — I’m not sure they have these receipts — APG passed it to songwriter Derrick Milano and others, that BTS began album work in Los Angeles around June 2025, and that the Netflix documentary “BTS: The Return” shows a song called “Swim” in development by August. As I said, pretty sturdy. And even if it were less sturdy, enough similarity and originality can lead to an inference of access. Somebody like me reaches the conclusion that, “these similarities are the sort that make coincidence super duper unlikely.” Here, the custody chain story seems solid.
The melody
I’ll get around to writing out the music and I’ll update this, but it’s a busy week, so for now I’ll just describe the transcriptions and take them as an accurate representation of the plaintiffs’ work. In short, they both contain the word “swim” a few times. And the single syllable “swim” is melismatically split across two pitches (redundancy intentional). But in the BTS song, Swim occurs on beat two and beat four. In the plaintiffs’ “Swim,” it occurs on beat one, and then it apparently echoes five times. I don’t know how prominent the echoes are, but a lot of the time, a musicological analysis might set the echoes aside to some degree and in this case, that might lead to the opinion that the composed melody in the plaintiffs’ song puts “Sw-im” on beat one, somewhat discounting the echoes, but here the four-beat resemblance depends on counting production echoes as melody. It further depicts and depends on echoes in the BTS recording that appear only near the end of the record. And again, how the plaintiffs’ demo treats its echoes, and whether they appear throughout, we cannot know today.
If you count the figure mechanically and look for strings of notes in common, you find none apart from a string of four consecutive eighth-notes on one pitch, “Do” as in “a deer a female deer.” And if you even look across the whole melody shown, looking for same pitches occurring at the same rhythmic placement, out of twenty-two notes, they’re the same exactly twice. Both matches land, again, on “Do,” the tonic, the most probable note in the key. The displayed excerpt also runs an odd two and a half bars. Melodic phrases generally come in twos, fours, and eights. I’d be very surprised if the demo doesn’t. BTS’s certainly does. Such an example that stops mid-phrase leads me to believe it was thoughtfully chosen to stop where it stopped.
The harmony
Think again about what “Do” means to you. The complaint leans hard on the two songs’ “unusual harmonies and their lack of tonal center.” Both float on two chords, neither the tonic. In BTS’s “Swim” the two chords are Dmaj7 and C#m7. In this truncated ten-beat-long example, C#m7 appears for only one beat in the plaintiffs’ work.
But let’s imagine these two chords appear adjacently in both works. There’s nothing unusual about that or about the chords themselves. The overwhelmingly available interpretation hears Dmaj7 as IV and C#m7 as iii in A major. I don’t find the pair particularly ambiguous or particularly “aqueous” as opined in the complaint, implying that it’s a device, therefore also stolen, which binds the unusual harmony and lack of tonal center to the watery themes of both works. Further, near the end of the BTS track, a repeated ii, I, V figure states A major outright. The tonal center I would’ve defaulted to throughout shows up at the end.
More about the harmony, the ostinato
The complaint describes “underlying ostinatos or arpeggio-like figurations” said to be the musical foundation of both.
The BTS figure, across four bars this time, is D, F#, A, C#, played twice per bar in straight eighth notes. That spells a Dmaj7 chord bottom to top, the most innocuous way there is to play one. An arpeggiator in up mode easily produces this 16-note figure with nobody composing anything. In the subsequent two measures of the BTS C#m7 ostinato, 16 notes, here again, are the same just twice.
It’s not that there aren’t two chords here playing arpeggios. It’s that it’s neither original, nor much the same — like someone picking a guitar chord, but not picking it in the same way. It’s just a chord, broken up, sometimes just spelling the chord as in the first two measures here, other times finding a little inner voice line as in the latter two. That’s what a guitarist or keyboardist provides on demand, spontaneously — and it’s a very available argument against copying.
I’m miffed at the red notes
Comparative exhibits color matching notes red; we musicologists do this. Red should mean duplication, with maybe a certain amount of leeway. The objective is to show and to teach, and thereby persuade. But what when the logic is opportunistic? On the one hand, kudos, if you’re an attorney framing the argument. On the other, someone like me is gonna push back. When you cover your transcription in red note heads, you scare the initial reader. But I look at it and see fewer than one in ten aligning vertically (occurring at the same time in the other melody) and I see begging, petitio principii if you’re feeling fancy. The analysis assumes and insists upon the copying, then recasts differences as a revision of the thing it was supposed to prove. Less fancily, “all the notes that match are copying, and the notes that don’t match are still copying except the notes were obviously moved.” Any skeptical eye dismisses that as unscientific.
But now seems a good time to remind everyone, I’m making a lot of inferences. I’ve never heard the plaintiffs’ work. The musicological allegations read to me like a piled-up mound of things that are true or at least plausible but not obviously consequential. Every example in this complaint requires some rope so I’m left skeptical on both the similarity and the originality legs.
And I’ll show not tell on this red notes stuff if and when I get my mitts on the plaintiffs’ work, transcribe it and color up my own red notes. I suspect I’ll be working in shades of pink. Or perhaps I’ll sidle right up beside the complaint. We can’t know.
Register your work!
A quick but informative footnote: the plaintiffs’ “SWIM” was registered in April 2026, after the plaintiffs had heard “Arirang.” Copyright is intact when a work is in fixed form. But enforcing copyright requires it be registered. This is a good example of the mechanics. For an unpublished work, late registration (after the infringement) takes statutory damages and attorney’s fees off the table. Statutory damages run from $750 to $30,000 per work, and up to $150,000 when the infringement is willful. Avoiding that tier is one of the benefits of pre-release originality clearance, which, in addition to preventing claims in the first place, demonstrates willfulness to avoid infringement.
My expectation is that this probably goes away.
But very humbly held for obvious and already stated reasons.
I expect the complaint puts its best foot forward. Its architecture leans on access, and Musicologize makes a lot of the unfortunate fact that in the Ninth Circuit that leaning could pay off through the inverse ratio rule, which lowered the bar for similarity as the showing of access grew. The en banc court abolished that rule in Skidmore v. Led Zeppelin in 2020. The Disco.ac logs can establish that people heard the file. After Skidmore, no amount of hearing it discounts what the music itself must show.
Which leads to another interesting question, not about the music, but the morality around ideas, expression, and co-authorship. I’ve known famous pro musicians who’d say, “If you were in the room, you helped write the song. And you’re on the credits.” Songwriting-room custom involves being present with an idea. Whether “they would have gotten credit in the room” differs from “the room liked their ideas but didn’t significantly regurgitate them” is the question. If access is a given, even if the writers of BTS did know the plaintiffs’ Swim, did they reuse anything that is the plaintiffs’ own?
My next expectation is that the likely fallback is selection and arrangement, the argument that unprotectable elements combine into protectable expression. Satava v. Lowry requires the elements to be numerous and their arrangement original before that works. A shared one-word title, a shared image of watching someone swim, a two-chord diatonic loop, and a root-position arpeggio make a short list. But there’s a ridiculous Reggaeton case also in the news that gets a lot of mileage from its double- and triple-counting accounting.
The complaint says:
In early 2025, Plaintiffs, all three accomplished songwriters and musicians, began
writing and composing a musical composition comprised of lyrics and music, by exchanging
ideas, music cues, information, and other data. In doing so, Plaintiffs meticulously selected
and arranged their melodies, lyrics, harmonies, rhythms, and other elements to create a
distinctly new, protectable musical composition.
“Meticulously selected and arranged.”
