Are We Doing Copyright Wrong in Music?

On today's podcast we have a truly amazing talk with Damien Riehl, a musician and technology lawyer who is one of the masterminds behind All the Music Project; an algorithmically generated catalogue of 471 billion melodies, which was then put into the public domain in an attempt to protect musicians from being sued for copyright claims. It is a fascinating conversation.

Damien’s Ted Talk here

News!

Gen Z is embracing a digital detox and the ‘Martha Stewart summer’

Apple Intelligence Is Gambling on Privacy as a Killer Feature

The iPhone is your karaoke mic in tvOS 26's Apple Music Sing

Nothing will stand in the way of independents (and why the world’s first three-person indie unicorn is only a matter of time away)

Looking for Rock Paper Scanner, the newsletter of music tech news curated by the Rock Paper Scissors PR team? Subscribe here to get it in your inbox every Friday!

Join the Music Tectonics team and top innovators by the beach for the best music tech event of the year: the Music Tectonics Conference, November 4-6, 2025 in Santa Monica, California!

Episode Transcript

Machine transcribed

Dmitri: [00:00:00] Damien Riehl is a musician with a bachelor's degree in vocal music and instrumental music, who's also a technology lawyer who clerked for state and federal judges and spent 15 years as a litigator, as a VP of VL. Damien currently uses generative AI to extract and generate valuable insights from a data set of 1 billion legal documents.

That's cases, statutes, and regulations worldwide. With his All the Music Project, Damien copyrighted 471 billion Mel Melodies, putting them in the public domain, seeking to protect you. Stole my copyright defendants who were wrongfully sued, which makes him the perfect person to unpack what's going on with music, copyright AI training data, and the case for and against fair use.

Welcome to the show, Damien.

Damien: Thanks so much, Dmitri I'm a huge fan of your show. I listen all the time, so it's really a thrill to be on with you.

Dmitri: That is amazing. I love that. I appreciate you listening and being here. So let's dive in. We've got a lot to talk about. What was this copyright mid [00:01:00] melody composition project that you and Noah Rubin laUn-hed in 2020? And why did you do it?

Damien: Sure. what we've done is we've copyrighted 471 billion melodies, and we did that through brute force. You can brute force a password by going A-A-A-A-A-B, a C. We did that with music where we went me until we mathematically exhausted every melody that's ever been and every melody that ever can be to the tune of 471 billion melodies that we then wrote to disc.

And under the burn convention, once they're written to dis, they're copyrighted. So I copyrighted 471 billion melodies and then I put everything in the public domain. To protect you stole my melody lawsuit defendants. so before my TED Talk, which has been seen about 2 million times, every defendant in those cases lost.

But after my TED Talk, which has again been seen 2 million times, every defendant has used my arguments and has won. So, Katie Perry, led Zeppelin for Stairway to Heaven, ed Sheeran twice, once in the us, once in the uk. All of them argued what I argued in my TED Talk saying maybe those melodies [00:02:00] are unoriginal, therefore, on copyright.

So that's, uh, that's my project. And, uh, we did that. you'd asked, why did we do it, mostly to be able to try to justify, what it was an unjust result. and that's, you might have heard about the Beatles. George Harrison had a case, my sweet Lord, that he was sued by the chiffons.

So the, you had my sweet Lord versus he's so fine. and what happened was George Harrison said, Hey, I've never heard the chiffon song in my life. And the judge said, I believe you that you don't remember, hearing the song, but what I think you did was subconsciously infringe. And I thought, as a law student, when I heard that, I thought, that's ridiculous.

I said, because every other case, plaintiff, somebody suing has to prove they have the burden of proof to show that the defendant did something. But in this case, the defendant, George Harrison, had to prove a negative that he didn't do something. And philosophically, it's impossible to prove a negative.

You can't prove that you've never heard a song in the supermarket, loudspeakers, or you've never heard a friend hold up their phone saying, Hey, listen to the song. It's [00:03:00] impossible to prove you've never heard a song. So I thought this is some a wrong in the law, and I thought maybe this, All the Music Project might be able to write that wrong, to be able to maybe help those cases.

And thus far it seems like it has.

Dmitri: are there many cases where somebody goes and does something like this? Puts everything out in Creative Commons to shift how people think about liability in this kind of copyright situation. or really undoing, undoing how somebody was interpreting a law by just creating this huge base of, creative commons content.

Damien: I don't know of anybody else. And I, maybe it's just my weird background where my bachelor's degree is in music. I was gonna be a choir director. and until two of my tenors started pUn-hing each other in the face, in the middle of the choir and I thought, oh, maybe I should go to law school. So there aren't a lot of music majors turned, lawyers turned technologists, and maybe it's the confluence of those three things that maybe gave me that idea to maybe try to right the wrong in the way that I have.

Dmitri: So you mentioned your TED Talk. I noticed that back then in that talk you said under copyright law numbers or facts and [00:04:00] under copyright law, facts either have thin copyright. Almost no copyright or no copyright at all. So maybe if these numbers have existed since the beginning of time and we're just plucking them out, maybe melodies are just math, which is just facts, which is not copyrightable.

And I was gonna ask you, did your experiment work? And you've already told us that it has, you gave it as examples. at least correlated examples

Damien: That's right. Correlation is not causation, but there's really good correlation.

Dmitri: so you're showing correlation that maybe this experiment worked, that has changed how these, suits are getting prosecuted, and decided on the other hand, have any new songs been created with obvious derivative compositions with the thought that your creative commons works, protects them.

Damien: That's, that's the right question to ask. and there's two aspects of what we've done. There's the melodies that we've copyrighted. there's the hard drives that we have, and then there's the idea behind what we've done. Really number two, the idea is the most important part of aspect. The hard drives.

Of course, they're cool. The fact that there's a hard drive sitting over here on my [00:05:00] shelf that has 200 million, or actually 471 billion melodies. It's cool that exists. But even more important is that judges have seen my TED Talk and clerks have seen my TED Talk and uh, and lawyers that are representing the defendants have seen my Ted Talk.

And so really this, uh, this idea,if I were one of those lawyers, I wouldn't cite some jerk that had a TED talk and I wouldn't say, Hey, this guy had a really good idea. But what I would do is I would take those ideas to say, your Honor, these are unoriginal melodies. and for an example, the Ka Perry case, that happened, the melody at that case was. ridiculously stupid melody, the jury found $2.8 million jury verdict over that ridiculously stupid melody that actually shows up in my dataset 8,128 times. So the plaintiff in that case was seeking to have a monopoly on that stupid melody for his life, life of the author, plus 70 years for something.

My machine crapped out at 300,000 melodies per second. And so the [00:06:00] real question is, should anyone have a monopoly? On that ridiculously stupid melody, or should that melody be unoriginal, therefore un copyrightable. And I would say to all of our musician friends, and I, I count myself among them, when was the last time you felt like, copyright vindicated you? That is, that you're actually protected by copyright? And now think about how many times are you worried about copyright that somebody's gonna sue you over copyright? and, because it's almost always used as a sword against songwriters. And almost, never as a shield to protect songwriters. I think musicians should really think about the value of copyright or,is there a value to you stole my melody lawsuit.

cases. Are you ever gonna sue somebody over stealing your, melody? If the answer's no, my, my project's probably protecting you more than it's, doing anything bad to you.

Dmitri: Okay, let me ask this a whole different way then. And be a little more direct. What's your position on whether successful artists should be able to copyright their songs? [00:07:00] And is it okay for someone to copy riffs from a famous song and then profit off of it without attributing and paying the original songwriter?

Damien: I would say that no, that, that is not right. and I would say that yeah, you should, if someone steals an entire song, of course sue them for all they're worth. Right? But I would say that, there's a difference between stealing the song and stealing just the chords. From a song because chords are not copyrightable courts have said that, right?

There are only four chords in the truth, right? so you can't really copyright a chord structure. You also cannot copyright a drum beat. courts have said that there are only so many ways to beat the drums,

Dmitri: The drummers always get screwed on this copyright

Damien: it's true. And you also can't copyright say the blues, right? Uh, these, so there, there is the elements of the song, that includes the lyrics.

And the, uh, melody and the harmony, the chord structure, the tambor of which are played, right? All of those things together create the quote unquote song. And so if someone steals all of those things together as the song, of course, they should be sued. But really, these, you stole my [00:08:00] melody. Lawsuits are just taking one layer of that, and it's not even taking the full melody from it, it's just taking a snippet of a melody usually from it.

And the question is, is that snippet standing alone, copyrightable or not? And maybe if it's the riff from, uh, you know, from a famous Rolling Stone song, maybe, right? But maybe that riff has existed in, in all of history, like it is exists in, uh, in the ether.

And that songwriter just plucked it out of the ether, just like some other songwriter, plucked it out of the ether. So the question is if two people independently came up with that same riff,

Dmitri: monkey.

Damien: That's right. But, yeah. And there's, in, in the law they call that independent creation. So if you come up with a riff on your own and I come up with riff on my own and we've never heard each other, then we can each have copyright in that riff.

So really what we're trying to do is that yes, if you're stealing the whole song to go to your question, yeah, sue them for all their way. But if it's just a riff or if it's just a snippet of a melody. The whole point of my thing is that maybe we shouldn't be suing [00:09:00] each other, all of these things. Maybe we should just make our own music because your song, might song, it'd be totally different than my song, even though we share the same melody.

An example of that is, blah blah, black sheep have any wool versus A, B, C, D, E, F, G versus, twinkle, twinkle little star. Those share the same melody. But until I told you that, you may not have ever

Dmitri: You just ruined my childhood.

Damien: Right. It's true because, and, uh, you might say, gosh, I had no idea that those shared the same melody.

And you know why? 'cause they're different effing songs, right? They, they, even though they share the same melody, they have different song space. And so nobody should be able to sue each other over that melody. So maybe we should think of melodies less as our intellectual property and more is part of the common good that we're all drawing from the same common wellspring.

Dmitri: Man, you should do a TED Talk.

Damien: I should maybe a second one.

Dmitri: So I would take the answer to my first question there. What's your position on whether successful artists should be able to copyright their songs? You didn't really answer, but what

Damien: Uh, yes, you should.

Dmitri: Yeah, it's just the rifts, that's the issue, [00:10:00] not the overall song. Right? If somebody copies it much more closely than clearly, they're really stealing versus the rifts.

The melodic rifts within songs are kind of like saying, you played an A

Damien: or you played a b.

Right. Or you played a, you know, a g chord.

Dmitri: Yeah. Yeah. Cool. Okay, so now we get to the real reason I wanted to interview you, which is 'cause I wanted to see what you thought about what's going on with copyright protections under the current government and to discuss the threat of generative AI on the careers of musical artists.

So let's start with copyright. Do you believe that copyright's being threatened and that the music industry is about to face an existential threat larger than what happened with peer-to-peer sharing and piracy?

Damien: I, I think that is the right question to ask and,Yogi Berra said, predicting is really hard, especially about the future. but let's, let's do some predictions. Right.so I, I think that, yeah, I think that we are really at an existential point as musicians and as copyright lawyers.

Something that we thought was [00:11:00] in the exclusive human domain that is creativity, quote unquote creativity. it turns out that large language models are able to crank out stuff that sounds pretty human and things that are quite quote unquote creative, autonomously. And so the real question is, should we be able to give the machines copyright in what they're cranking out?

that is for my 471 billion melodies, should I have copyright in all 471 billion of those melodies? and I would say probably not, like my machine shouldn't have copyright in those 471 billion melodies because if I did get copyrighted in it, then nobody else would be able to make any more music.

Right. And so this is, uh, I, I think that the US copyright office is correct to say that if machine generated then not copyrightable, because otherwise if we don't have that ruling, then the machines will essentially edge out and squeeze out all the humans, and there just won't be any more music to make.

The machines will just prove force. So number one, maybe machine created works are Un- copyrightable. [00:12:00] So then thing number two is what if I'm a mach, a human that is jamming with the machine

Dmitri: Hmm.

Damien: able to create this output? And what the US copyright office has said, okay, Damien, if you're jamming with the machine then you get copyright in what you as a human have created, but you don't get copyright in what the machine has created.

That's probably the right move. Right? Okay. So like, uh, so if the machine cranks out a bUn-h of lyrics and a bUn-h of melodies and a bUn-h of harmonies, and then I maybe tweak a little bit of those, I get the tweaks, but I don't get the bass song, as a, as an

Dmitri: Hmm.

Damien: But the tricky part is how do you un bake that cake, right?

what if the machine cranks out, 30, different notes and I tweak one of those notes, or three, five of those notes every other note I tweak. Do I get copyright in every other note? That doesn't make any sense. And if Microsoft Word, you have track changes, right? So you could be able to say, if you track changes, if it cranks out a book and you change 25 words that are spread throughout that book, do you get copyright in those individual words?[00:13:00]

And music doesn't have track changes. So how,if push comes to shove and that musician is interviewed, and the copyright off says what part was human created and what part was machine created? A musician's gonna say, man, I was up till 3:00 AM. I don't, I have no idea what the machine created and what I did.

I just jammed with the machine. So really the real question is, uh, you asked, is copyright being threatened? I would say it's maybe threatened in the sense that, what we call joint works, that is if Dmitri and I have a, joint work together, historically, you would have, you and I would have a, a whole, we would each have, 50% of the whole.

So that uh, when we sell the book, we get 50% of the proceeds. Even if you made 90% of the work and I made 10% of the work. And, uh, we do this because the courts aren't gonna wanna argue, or hear the argument from me to say, no, my 10% was the most important 10% and your 90% was just dribble. Right? They wanna say, no, forget it.

Dmitri gets 50% and I get 50%. That's 'cause they don't want to unbaked the cake, right? But the problem is how do you unbaked the cake [00:14:00] of music? How much was machine created? Maybe the machine created 90% and the human created 10%. Are courts gonna wanna unbaked the cake or will they be able to unbaked the cake that way?

So I, I would say that yes, copyright is threatened because if machine generated then on copyrightable, as we as musicians jam more with the machines, maybe more and more of our work is gonna be on copywriter.

Dmitri: And what about this existential threat for the music industry? is that to go even further into the future once that happens? do you think that? It gets to a point where music can't be considered valuable because of,the impact it has on music, creativity, and the ability to sell access to music.

I.

Damien: I, I think that the music, industry is not going anywhere. The reason I, I think that is that, we've had music, since, ca prehistoric times, right? Sitting around campfires. we would make music. And so there's something primal about music making and music listening that I think is not going to go away.

the real question is, who's gonna make [00:15:00] money from

Dmitri: Right, right.

Damien: And, and that's something that you and I are part of the folk group, where that's, we have some of the smartest people in the business, of music, thinking through how people are gonna make money with music.

And the question is,is that music, money going to accrue to the musicians? Or is it gonna accrue to the business people that are using machines to crank out cheap music without musicians? And I think that, that future is hard to predict. I hope musicians win. I hope the humans win.

Dmitri: I wanna get more into some of the political context. we need to take a break, you mentioned the F group, PHO. It's the name of a Vietnamese dish, a noodle,it's an online discussion group that both Damian and I are a part of that you can look up and join as well.

Well. Um, but yeah, I just wanna make sure folks know what you were talking about there. Started by Jim Griffin, who still facilitates that conversation and a couple other folks. Um, they originally met over fa at a restaurant in LA and then it turned into this multi-decade online conversation about music and the law and rights and the future and so forth.

Damien: Yeah. And then something about [00:16:00] Jim, Jim Griffin, if you don't know him, he was responsible for the first downloaded MP three of a song that, he was part of a record label that said, Hey, let's have an MP three download of, this Aerosmith song. so anyway, so that's Jim Griffin's, had the first MP three, and he is led the Pho group.

Dmitri: And we've had him on the podcast as well. You can go back and look up the uh, Jim Griffin interviews as well. Okay. When we come back, I want to ask you a little bit more about the political context and generative ai. We'll be right back. Okay, we're back. Damien. This is a blast. I'm having my mind blown as I do every time I see you run into, you talk to you, read emails from you, et cetera, which is a blast. Let's get a little into this current litigation around generative ai. You know, the major labels. have initiated a lawsuit against Suno and Udio for training their data, on unlicensed music.

Then recently there was news that the labels were negotiating with those companies as well. How does the current political conversation around copyright, because [00:17:00] there, I mean, the other piece of this news is. there's been some changes at the copyright office. how do, how does this political conversation around copyright impact these debates and specifically the litigation related to generative AI music platforms?

Damien: Sure. so I'm gonna, before I answer the question, uh, and this relates to it, I'm gonna touch back to what you and I discussed earlier, is that, listeners, think about the last time you felt like you were protected by copyright. And I would say that most listeners are gonna say, I've, I've never felt protected by copyright.

And then, so that's question number one. Question number two is, what is the purpose of copyright? And the purpose of copyright is to exclude others. That is if somebody kind of treads on your, your property, your copyright, then you can sue them. So you're excluding them from your copyrighted works.

And then num question number three is if you never sue somebody over copyright infringement, does copyright even matter? That is if you're not gonna sue somebody for stepping on your copyright because either you can't afford a lawyer to sue them or because you just, uh, don't [00:18:00] know even where to begin.

Then how much is copyright really protecting you? And so that's, those are the three questions I wanna ask, before I answer your ask, about the political conversation. now when you think about, okay, there's, uh, SUNO and Udio, they are being alleged to have,taken a whole bUn-h of recorded music and being able to essentially extract the idea of the blues and the idea of classical music.

And then, uh, once it's extracts those ideas, then it can create new music like that. And in copyright, there's what's called the idea expression dichotomy ideas are Un- copyrightable. Okay? Only the human made expression of those ideas is copyrightable. Okay? if you think about,think about this plotted on two dimensions.

So you have the x axis and the y axis. So the x axis going left and right, and the Y axis going up and down. You can plot, most words in the x axis and y axis. So you can plot guitar on the x axis. You can plot, the violin on the X axis. now you can add a third dimension, a Z axis. So now you have three [00:19:00] dimensions, and you can actually plot words and concepts in the three dimensions.

Okay, now plot it on fourth dimension, and it's impossible for our brains to even figure out what a fourth dimension looks like right now. Do a do a fifth dimension. Do a 10th dimension. Do a 100th dimension, do a 1000th dimension, now do a 12000th dimension. And 12,000 dimensional vector space is where large language models and generative AI live.

So somewhere in that 12,000 dimensional vector space of un copyrightable ideas, somewhere in those 12,000 is the idea of Bob Dylan, this and the idea of Earnest Hemingways and the idea of Pablo Picassos. So these are all ideas, not expressions of those ideas. And again, ideas are Un- copyrightable. And the question is, can Bob Dylan sue you, Dmitri?

For sounding too much like Bob Dylan, and the answer is no. Because if he could do that, then he could have sued every singer songwriter since 1970 because everybody sounds like Bob Dylan. The idea of Bob Dylan, this is not [00:20:00] copyrightable. The idea of Pablo Picassos is not copyrightable. the idea of est hemingways is not copyrightable because if it were, you would have essentially a sword being struck you every single time you make a song.

Because everybody sounds like everybody else. That's the idea of the blues. Or the idea of hip hop or the idea of EDM, right? Ideas are Un- copyrightable only the human made expressions of those ideas are copyrightable. So getting back to your question about Suno and Udio, if I were representing Studio and Udio, I would say that, listen, all they were doing is extracting those Un- copyrightable ideas, extracting the Un- copyrightable idea of Bob Dylan, this and Earnest Taming willingness, and then creating machine created outputs that again, as we talked about earlier, are Un- copyrightable.

So I know that this answer is not gonna be satisfying to many of your musician listeners, and some of you might be upset by what I'm about to what I've just been saying. But I would get back to, are you really making any money from copyright anyway? Or, are you, do you feel more threatened by copyright [00:21:00] than helped by copyright?

And when you think about selling records, is this copyright that's selling, helping you sell the records, or is it the music licensing collective that is a separate federal statute. That enables you to get paid for your recordings because it's the statute really that helps you get paid. And that has nothing to do with copyright.

It has everything to do with the federal government saying, yes, we need to get musicians paid for their work, not copyright, but under the MLC Music Licensing Collective

Dmitri: A mechanical licensing

Damien: and Mechanical License Collective. so the real question is copyright even doing the heavy lifting here? And the answer is probably not, because if the courts were to say, yes, ideas are copyrightable.

Then you might get sued by Bob Dylan or somebody else. I think we want the courts to say no. The idea of a genre or the idea of a particular style of music or a person style is not copyrightable, and that's actually good, and it doesn't actually get you paid anyway. It's the recording that gets you [00:22:00] paid.

Dmitri: Yes. You're blowing my mind, Damian. It's true. So let me ask you, do you think the fair use argument that AI companies can train on unlicensed music training data, has a legal standing.

Damien: yeah. So the, when you think about any copyright infringement question number one is the thing copyrightable in the first place? And if the answer is no, do not pass, go, do not collect $200. You don't even have to get to the fair use question. So question number one is a copyrightable in the first place?

If no, then analysis done. Question number two, did Dmitri have access to Damien's song? To show that Dmitri copied Damien's song, if no access, then no copyright infringement. 'cause you can independently create your thing or not. If the answer is yes, Dmitri had access to my songs, then you would go on to question number three, which is fair use. But I would say that most people jump right to fair use without thinking about question number one, about is it copyrightable in the first place, or number two, did you have access to that particular song? I would say if you make it through those first two gates, right, without passing, go, without collecting [00:23:00] $200, then the question of fair use is there are several factors of fair use, including how much of the original did you take and, what's the nature of the output is another question.

and how close is the output to what the original was? And so I would say that, uh, there's a pretty good, fair use argument that the AI companies have to say, number one, I extracted those Un- copyrightable ideas. Not the expressions of those ideas. And then I took those Un- copyrightable ideas and I recombined them into Un- copyrightable outputs.

So that transformation, that's something that the courts talk about is the use transformative. Taking an Un- copyright, Un- copyrightable idea and transforming, it seems pretty transforming into, a new one. So I would say that's a long answer to your question, but number one, is it copyrightable?

Maybe not. Number two, did they have access to your particular song? Maybe not. And then only then number three, is it fair use? And the answer is maybe it's pretty transformative.

Dmitri: Well, but they do have access, right? If they trained on, the songs of all the [00:24:00] major labels they had access to it, they might have been extracting Un- copyrightable components, the Bob Dylan ness. Maybe they were extracting the Bob Dylan ness or the, what, whatever artist ness you want, want to say.

But didn't they have access to it in the process? Weren't they using that to extract

Damien: Maybe, maybe if you're Bob Dylan, but what if you are, our listener is a independent singer songwriter. Maybe they didn't train on your data. Maybe they, you're just, like a folk singer and they just trained on a whole bUn-h of folk songs and so they no more trained on your data any more than you trained on Bob Dylan.

Dmitri: So are you saying that they might have trained on a lot of stuff that sounded like the content from the major labels and then made it sound that way, that they actually could make the argument that they never uploaded, cop copyrighted songs to their training models.

Damien: Yeah, they, they could maybe argue that your particular song, they didn't, do, So they, you can imagine, maybe they trained on a bUn-h of things, but not on your particular song. and another [00:25:00] argument is that, yeah, we've trained on your data, but for, for a copy to be made for copyright infringement, there has to be a true copy both on the input and the output. And the question is, if I've trained on your data, but the output is not a quote unquote copy. Of your input song, is there any copy made at all?

Dmitri: So you're saying that in order for the major labels to win, they would have to demonstrate that a copy was made in addition to the fact that they trained on the, on the, their songs.

Damien: right. Copyright is literally the right to copy, and if there's no copy, then there's no copyright infringement.

Dmitri: This is really interesting, especially coming from a musician like you.

Damien: and I would say that, uh, and, uh, people, often ask me, after your, All the Music Project, did you get any death threats or anything like that? And I, I would say no. I almost universally, people were very supportive. Musicians were supportive of what I've done because they, musicians have told me, I feel like I have a target on my back every time I write a song.

But I feel like your, All the Music Project has taken that target off of my back to try to protect you. And,I worry a [00:26:00] little bit that what I'm talking about right now, it seems like, uh, uh, it's gonna have, targets put on my front. that is people are, are maybe gonna be angry with me, but I would go back to when was the last time copyright protected you? That is when is the last time you were thinking about, suing somebody over copyright infringement and when was the last time you actually sued somebody over the copyright bridge? And so I would say if the answer is I've never felt protected by copyright and I've never thought about suing anybody, then copyright is not really doing you any good. What's doing you good is going out there on the road and building an audiencemaking sure that people love you. And maybe it's not the copyright, the thing that has, been making you money all these years. Maybe it's your personality, maybe it's your story. And as machines make more and more music that are maybe background music, they're still not gonna have the story of Taylor Swift. Which is the reason that people really follow Taylor Swift. Yes. It's her music's good, right? People think, but it's the story of her. It's the fact that she's dating a football player, right? The machines aren't gonna be able to do that. and the machines, machines do better than us at chess.

but we don't watch machines play each other in [00:27:00] chess because there's no story there. So I would say musicians, you should really think about what draws your audience. And is it the copyright in your songs? No. It's the fact that you've built this relationship with your audience. What's making money, not the copyright?

Dmitri: Okay. I was just gonna ask you, so it sounds like you're not really worried about the threat of AI music changing the business model of, of music for artists and maybe for record labels as a result.

Damien: I wouldn't say, I'm not worried about that. I am worried about that. And the reason I'm worried about it's because, uh, you can imagine if you're a large streaming platform, that has to pay out money to human artists, but you don't have to pay out money to machine artists. there's an incentive for you as a streaming platform to play more artificially generated music that is not copyrightable, that I don't have to pay for.

So I, I do worry about, for background music and stuff that is just, happens to be on, uh, that more and more of that is gonna be machine created and less of that is gonna be paid to humans. So I worry a lot about that. And I also worry about, my friend, will Page, who's [00:28:00] the former chief economist of Spotify, also another Pho member.

he talks a lot about paying money, paying attention to something. that it's literally paying like money, attention to something and, that people are paying attention, to streaming, Netflix to Hulu and to all these, uh, YouTube, right? So as a, we as humans can only listen to one thing at a time.

we could either listen to a TV show or watch a TV show, or listen to a podcast or listen to music. And what I do worry about is the percentage of music listening has shrunk. people are paying more attention to other things. And then that shrinking music, attention is then being diluted further by machines taking more of that money.

So I do worry a lot about that, that, that musicians, and music as a part of the social zeitgeist is shrinking. And if that happens, I think that we will societally have a huge, massive loss. I think that since caveman days, we've been,nourished by music. And if that goes away, if people are not paying attention to music, that's an existential threat that we should [00:29:00] avoid.

Dmitri: But all those things that they're getting distracted by, whether it's Netflix or gaming or whatever, are almost always accompanied by music as well. so I guess the trick is just to make sure there's still human creators behind that music that's in the background of our lives.

Damien: that's right. and the question is, how much do the game creators and the TV show creators, how much is the cheap cost of machine created thing? how much is that gonna be a siren song to, if you, excuse pun, to not have to pay a human? if I'm a game creator and I could just crap out something with pseudo Udio without having to pay the human stuff.

And I, I know that the suno and Udio thing is not copyrightable because machine created things are Un- copyrightable that shrunk cost and no copyright royalties for the rest of the duration of that game. I worry that that cost cutting and,incentivizing, will go away from humans.

Dmitri: Yeah, I don't know, man. Music is cool in a way that human created music is cool in a way. It'd be hard to tap into some of the cultural kind of, specialness, that, that gets tapped for syn and that [00:30:00] sort of thing with quote, background music. I mean, the name background music alone kind of says it all.

Damien: Uh, right, right. It's true. and I, as a musician, I agree. like there's, there are few things more, exhilarating than being at a live concert. and, uh, we, we need to have that liveness. my worry is that business people don't feel the same way that you and I feel.

Dmitri: yeah. Alright, let's tie this all together. How is the fair use argument by generative AI companies related to your, your, All the Music Project, that you did back in 2020? like, we've gone through a series of iterations of this conversation, but do you see a relationship between those two?

Damien: Yeah. and it goes back to what we talked about earlier. Question number one, is it copyrightable in the first place? And the answer is,under my, All the Music Project, if, uh, if it's in my mail, the music project, then maybe it's not copyrightable in the first place.

Number one, maybe not copyrightable. Number two, did you have access? And then number three is a fair use. So there's a tangential relationship that people jump right, to fair use. And they also, uh, jump right to substantial similarity. this is something we haven't yet talked about [00:31:00] in, in our, uh, group.

That is if your melody is substantially similar to my melody, then we would be able to maybe have a copyright infringement claim. but the problem is, uh, you know, are we also substantially similar to a third party? or are we also substantially send substantially similar to a Bach piece?

Or a Mozart piece or a Hiden piece. and so the real question is, substantial similarity. Yeah. if you're an EDM producer or if you're a singer songwriter, of course you're gonna be substantially similar to other people in your genre. so anyway,this all gets back to,they've said since biblical times that, uh, there's nothing new under the sun.

That, and said another way, we all stand on the shoulders of giants. And that we're all, Rick Rubin, who your listeners probably know, was interviewed by Rick Beto and Rick said, is it hard to be creative? And Rick Rubin said, Hey man, everything's been done before. everything has already been done.

So the creativity is taking two different things that have already been done before and mixing them together. So in his case, hip hop and rock music, [00:32:00] putting it together, is the creativity. Rock music already existed, hip hop existed, but putting the two together, man, that's the creativity. So when we think about what we as artists are gonna be doing to be creative going forward, none of us creates anything new.

Everyone just combines old things and puts them together in a new way that our audience delighted. And the question is, is copyright helping you do those things? And I would say no. copyright's probably hurting you more than it's helping you, because if you're not suing, it's not helping you. If you're not suing somebody over over your songs, so maybe we should think less about copyright and just think about more about let's combine two disparate things that maybe combine those things in a way that's quote creative before the machines get to it first.

Dmitri: I always love that, that reminds me of, uh, Clayton Christensen's, the Innovator's Dilemma that says innovation. It's exactly what you quoted Rick Rubin on. It's like there's nothing new. It's just taking seemingly unrelated things and putting together, that's what creates innovation. And I guess you're saying the same thing[00:33:00]

Damien: That's right. and when you think about that's, that's the nature of liberal arts. I was a liberal arts major and the whole point of liberal arts is to take two disparate concepts and bring them together. And in a sense, that's what my All the Music Project did. So I was a musician, number one, I was also a lawyer.

Number two, I was also a technologist, number three. And I took all those three disparate concepts and I kind of brought them together to say, oh, here's an idea. Nobody's ever done this before. So I think that as we work side by side with machines in our large language model present, we have to think about what are the disparate concepts that we as humans can put together that the machines cannot yet put those disparate concepts together.

That's where we should be running. That's where you, an artist, a musical artist should be running. That's what visual artists should be doing. Putting together those disparate concepts in a way that the machines can't yet.

Dmitri: Well, this has been great, Damien. Is there anything else you wanna say about the conversation? I think that was a great summary right there. but uh, is there anything else you want to add before we wrap up for today?

Damien: I wanna, yeah. So I guess the, in, in closing, I would say that, I love musicians. You obviously you love music, [00:34:00] otherwise you wouldn't be listening to this podcast. I think we should get musicians paid. Let's get musicians paid through ai. and much like the mechanical licensing collective, MLC gets people paid in a way that's not copyright related.

It's just getting them paid. I think we should do the same thing for ai. That is, if AI uses your music to train, you should get paid. We should figure out how much, of your, uh, was taken, how much of it is in the output and just like the MLC pays out, I think AI should do that too. so I would say, in this is kind of a happy note to be able to say, yes, let's get musicians paid, but let's not rely on copyright to do it.

Let's rely on an MLC like way to do it. Because if we rely on copyright to do it, then you might get sued for being too Bob Dylan, like, or to sound like too much like this EDM person down the street. You don't want copyright to do the heavy lifting. You want a statute to be able to protect your rights, much like the MLC does.

copyright, uh, should be its own thing. and there is nothing new under the sun. It's really not your melody. so you [00:35:00] shouldn't feel like you should have a target on your back.

Dmitri: Amazing. Damien, this has been great. I've learned so much and I really appreciate you taking the time to go through all these iterations. You, you've performed some great aerobics in this, or acrobatics I should say. You've performed some great acrobatics in this conversation and I was actually able to keep up.

So thanks so much,

Damien: thanks so much for having me, Dmitri I love your show and I'm thrilled to be honest. So thank you so much for having me.

Let us know what you think! Tweet @MusicTectonics, find us on LinkedIn, Facebook and Instagram, or connect with podcast host Dmitri Vietze on LinkedIn, Twitter, and Facebook.

The Music Tectonics podcast goes beneath the surface of the music industry to explore how technology is changing the way business gets done. Weekly episodes include interviews with music tech movers & shakers, deep dives into seismic shifts, and more.