This is pretty cool -- Jason Isbell, David Lowery (Camper Van Beethoven, Cracker, UGA prof), sax player Eduardo Calle and Austin's Guy Forsyth have filed a class action lawsuit against Suno to force them to stop allowing users to generate music that evokes a specific artist's identity via AI.

AI class action lawsuit

Edited September 3Sep 3 by jimmyjazz

I saw this the other day, good shit

Suno denies the claim, saying their software strictly prohibits artist prompts. It turns out all one has to do (among other things) is put spaces between the letters in a name to get around that restriction; e.g., "J a s o n I s b e l l".

Edited September 3Sep 3 by jimmyjazz

I'm sure determining the metric or methodology would be difficult, but they should really make it such that music created using AI does not grant the "author" any kind of copyright. Like, currently ASCAP goes and collects money from radio and streaming and tv and concert venues and girl scout campfires and etc etc etc and then divvies up that money to the songwriters and publishers based on some mix of counting and detecting and extrapolating and black magic. There should be some congressional regulation and an organization set up to collect all of the different kinds of publishing revenue for this AI stuff and give the money to songwriters. It shouldn't be free to the people who are exploiting the music because it's derived from the creative work of humans, and it shouldn't go to the people who wrote "give me a song in the style of brandi carlile about a group of vigilante dogs who rescue a dinosaur island from a diamond hungry capitalist" into some prompt because obviously.

Paging @TwiceHorn for his insight.

22 minutes ago, Celery Man said:

I'm sure determining the metric or methodology would be difficult, but they should really make it such that music created using AI does not grant the "author" any kind of copyright. Like, currently ASCAP goes and collects money from radio and streaming and tv and concert venues and girl scout campfires and etc etc etc and then divvies up that money to the songwriters and publishers based on some mix of counting and detecting and extrapolating and black magic. There should be some congressional regulation and an organization set up to collect all of the different kinds of publishing revenue for this AI stuff and give the money to songwriters. It shouldn't be free to the people who are exploiting the music because it's derived from the creative work of humans, and it shouldn't go to the people who wrote "give me a song in the style of brandi carlile about a group of vigilante dogs who rescue a dinosaur island from a diamond hungry capitalist" into some prompt because obviously.

I'm inclined to agree, but I'm afraid the technology and user savvy will soon outrun the law/regulations. Perhaps there could be some kind of digital watermark associated with AI generated music but I don't think it would be that hard to craft a more generic prompt that basically gets the same job done.

Right now, they are actually using AI to detect AI. Most places have demonitized AI music as far as streamers go too I believe

1 hour ago, Celery Man said:

I'm sure determining the metric or methodology would be difficult, but they should really make it such that music created using AI does not grant the "author" any kind of copyright. Like, currently ASCAP goes and collects money from radio and streaming and tv and concert venues and girl scout campfires and etc etc etc and then divvies up that money to the songwriters and publishers based on some mix of counting and detecting and extrapolating and black magic. There should be some congressional regulation and an organization set up to collect all of the different kinds of publishing revenue for this AI stuff and give the money to songwriters. It shouldn't be free to the people who are exploiting the music because it's derived from the creative work of humans, and it shouldn't go to the people who wrote "give me a song in the style of brandi carlile about a group of vigilante dogs who rescue a dinosaur island from a diamond hungry capitalist" into some prompt because obviously.

As it stands, that is the case. Copyright, as statutorily enacted at least, requires a human author, and patents, as statutorily enacted, requires a human inventor.

I say as statutorily enacted, because at least theoretically, the statutes could be amended to permit non-human authors and inventors. The question remains open whether Article 1, Section 8 of the constitution requires human authors and inventors.

The relevant Constitutional provision:

To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries.

Somewhat interestingly, the human author requirement of the Copyright Act was not decided in the context of AI. Someone tried to copyright a "selfie" taken by an orangutan. No dice.

To infringe a copyright, the copyright work must be "copied."

Copying, in turn, is proven by circumstance of the infringer's "access" to the copyright work and a resulting "substantial similarity" between the copyright and infringing works.

In this infringement analysis, care has to be taken that the similarity occurs in original (to/by the author) elements of the work and not "ordinary," unoriginal elements. You see a lot of this arguing back and forth in music copyright cases: is it a standard chord progression or is it something original to that song by that author.

The amount of similarity can be pretty small, just a few bars of a song, if it is distinctive and clearly original to the author.

For those unmusical like myself, an analogy for this is that similarity in a literary work or movie must come from something more than broad plot elements, what are known in drama and literature, and copyright law as "Scenes a faire." Boy meets girl, boy loses girl, that kind of thing.

Most of the copyright litigation here to date has centered on training an LLM on copyright material without permission to do so. As a first matter, training an LLM would seem to require at least one duplicate or reproduction of a copyright work and that would be unlawful. For example, digitizing a copy of For Whom The Bell Tolls for input to the model.

So, LLMs may also have that problem because even a digital music file is probably duplicated at least once in the process of training the model.

But, the legal treatment of what an LLM does with the copyright work once trained is a blurrier question.

I think pretty clearly, outputting any recognizable portion of a copyright song would constitute infringement, just like it would in more ordinary music copyright cases.

Outputting a song "in the style of," would be far less clear if it avoided any similarity in the actual musical notes.

I don’t know what the status quo is in regards to AI generated music, practically speaking. Does the person who prompted the AI model hold the rights?

Obviously I am not a lawyer. Even for the generically created stuff that isn’t trying to copy the style of a specific artist, if I understand AI at all it’s kinda copying from everybody all at once? Obviously it is artistically meritless, but it sucks if tech just comes up with a way to cut songwriters out of the commercial exploitation of songs.

All that said.

This is not a copyright suit.

It's an NIL suit. All of the counts are for violations of various state's NIL and right of publicity statutes.

They do not contend that Suno violates any music copyrights, some of which Suno has apparently duly licensed.

They do contend that they are misappropriating elements of musicians' identities, their "look and feel."

Edited September 4Sep 4 by TwiceHorn

2 minutes ago, Celery Man said:

I don’t know what the status quo is in regards to AI generated music, practically speaking. Does the person who prompted the AI model hold the rights?

Obviously I am not a lawyer. Even for the generically created stuff that isn’t trying to copy the style of a specific artist, if I understand AI at all it’s kinda copying from everybody all at once? Obviously it is artistically meritless, but it sucks if tech just comes up with a way to cut songwriters out of the commercial exploitation of songs.

So this is an interesting question, also. The Copyright Office I think has provisional rules permitting registration of a copyright in a prompt or set of prompts and the output that results.

Currently, you can only copyright human added elements to a work that incorporates AI generated elements, which are excluded from the copyright claim. Apparently visual artists will edit and rearrange something that starts AI. That contribution is copyrightable, as I assume some human modification of AI music would be.

Edited September 4Sep 4 by TwiceHorn

3 minutes ago, TwiceHorn said:

All that said.

This is not a copyright suit.

It's an NIL suit. All of the counts are for violations of various state's NIL and right of publicity statutes.

That's the crux of the issue. Copyright discussions are irrelevant.

They do some interesting stuff with music rights - like, a songwriter has the right to first commercialization of any songs they write, but once they release the song anybody can cover it and can release a recording of it and they have to royalties to the songwriter but that amount is set by statue. And it is true that for performance rights, the performance rights organizations have different ways to count and get some idea of what songs are being played with what frequency in what capacity but it's a lot of extrapolating from their to distribute all of the royalties collected for song performances in the US to the hundreds of thousands of songwriters and publishers.

But yeah I derailed this post immediately.

3 minutes ago, jimmyjazz said:

That's the crux of the issue. Copyright discussions are irrelevant.

I don't think it really implicates the problems of human vs. AI. It's just going to boil down to is it similar enough?

NIL derives from trademark, and similarity is judged on the "likelihood of confusion" standard, that is, are consumers likely to be deceived that Suno's products are the same as, or sponsored or endorsed by, the original artist.

In trademark law, an intent to copy or duplicate another's mark can be taken as evidence of a likelihood of confusion. As is said "if we find an intention to copy and deceive, we may presume that the copyist succeeded." I don't know if that has been wholesale adopted in NIL law, but it seems likely that it might.

There are a few little wrinkles in the law(s) here and there, but that's kind of the gist of it.

Edited September 4Sep 4 by TwiceHorn

I wonder how true human creation "in the style of" a well-known artist might muddy the waters? If I can sing and write a song that an adept listener might think is Jason Isbell, did I appropriate his name, image, or likeness?

5 minutes ago, jimmyjazz said:

I wonder how true human creation "in the style of" a well-known artist might muddy the waters? If I can sing and write a song that an adept listener might think is Jason Isbell, did I appropriate his name, image, or likeness?

Yes. Or at least under the theory of this case.

Edited September 4Sep 4 by TwiceHorn

One thing in your more typical NIL case, kind of like athletes, the name, image, and/or likeness is typically infringed by using it like a trademark, to promote products.

This is somewhat unusual in that the appropriation of the NIL is the product itself. In that sense, it's kind of like EA Sports and the NCAA Football, etc. games.

One thing that occurs to me is that copyright law is highly preemptive. States have, from time to time, come up with various anti-copying schemes. Florida's "boat hull protection act" is something they've tried a couple of different times. It protects boat hull designs from copying/similarity.

And was ruled to be pre-empted by copyright law, because it was no different in proof and remedy than copyright law and purported to cover what copyrights might cover (although boat hulls are likely excluded as too functional).

Some attempts to use "common law" misappropriation theories, like a form of civil theft, have been used to attempt to protect software from functional imitation, that is similar or identical function and look and feel, but from different code. Also preempted by copyright law, again because the subject matter is that of copyright, the proof was the same (copying), and the remedy, injunction and lost profits, the same.

NIL tends to have that extra element of proof of similarity that leads to likelihood of confusion, otherwise it might be preempted, as well. If Lowery & Co try to stretch it too far, they might run into that.

So, who wants what Suno sells, anyway? What's the point?

1 hour ago, TwiceHorn said:

So, who wants what Suno sells, anyway? What's the point?

Seriously? If I could create a work highly reminiscent of (say) Paul McCartney and then sell it as being Jimmy's Jazz you don't think that would have massive value, at least above and beyond my meager skills? Who gets the royalties? Certainly not Paul McCartney.

10 hours ago, TwiceHorn said:

So, who wants what Suno sells, anyway? What's the point?

I don't know if this is all super germane to the legality being discussed, but I know a Nashville singer/songwriter who uses Suno for shoppable demos. He writes a song, makes a shitty phone recording of it, him playing the chords on acoustic guitar and singing, and has Suno take that raw recording (and his prompts) to build it out into a pretty damn nice production with drums, bass, piano, pedal steel, lead and backup vocals. Certainly a well enough made production to shop around his channels. He's basically using it to save the money he would have spent on the studio, recording engineer and studio musicians. He can obtain a similar result in far less time and cost this way.

Quite a few youtubers are using Suno to create the backing tracks for songs in their content. The gear demo video that starts or ends with the gear being demonstrated in the context of a throw away tune. Spend 5-6 hours making a backing track by hand to demonstrate you using the gear in question with, or 10 minutes in Suno. Its not real music, anyway.

And you have a good number of people with no ability to create music who get to feel the endorphin hit someone gets when they write a song. I think this is going to be used a lot for the crossover artist. The kid movie star with a following on social media, "creating" music for their fans. Capitalize on your brand made elsewhere to carry it over into the realm of music.

But I think the bigger issue is going to be when Suno creates a partnership with Spotify to give end users the ability to prompt their playlist with Suno generating the songs mostly on the fly in the style of X. In this case, Joe the music consumer says, "I love Jason Isbell and W.A.S.P., so craft me a playlist in a mashup of their styles. But I gotta tell you, I've got a fever, and the only cure... is more cowbell. So make sure every song features a ridiculous amount of cowbell." Hits go. Suno generates the music to fit the consumers desire, and Spotify delivers it to them. Suno gets paid something that undercuts what Jason Isbell and W.A.S.P. would have been paid, Spotify has lower costs, and thus the transfer of income from human hitmakers like Jason Isbell (and WASP) is made to Suno and Spotify investors/shareholders.

So I think this case is really about trying to get ahead of that. But I think it gets murky quick. Given this is a mashup of Isbell and W.A.S.P. and features a creative choice to include more cowbell, directives all given by the consumer, where does the NIL apply in a scenario like this? Further, AI is capable of making far more than direct connections mentioned by an identifying name. If someone wants to describe a Jason Isbell tune to Suno without mentioning him by name, just describing the characteristics of his voice, guitar playing, his songs, their arrangement and production choices, the theme of the song, the model will spit out something at least superficially resembling the Jason Isbell tune and will have used a training corpus including Jason Isbell tunes to generate it. It will take that input, use it to traverse its neural network to make musical choices similar to Jason Isbell and related artists to generate an output that sounds somewhat like Jason Isbell. Was his NIL used in this case? How could it be proved in a court of law?

Edited September 4Sep 4 by Goredho

The Isbell W.A.S.P. collab is definitely what we need

one of my favorite composers weighs in at parliament.

13 hours ago, jimmyjazz said:

Seriously? If I could create a work highly reminiscent of (say) Paul McCartney and then sell it as being Jimmy's Jazz you don't think that would have massive value, at least above and beyond my meager skills? Who gets the royalties? Certainly not Paul McCartney.

Is someone going to do this at commercial scale?

Or is it just for duffers that crave more Prince or Jim Croce?

I suppose that could drive subscriptions.

Just now, TwiceHorn said:

Is someone going to do this at commercial scale?

Or is it just for duffers that crave more Prince or Jim Croce?

I suppose that could drive subscriptions.

I very much fear that it will be used to replace actual artists on a massive scale.

45 minutes ago, jimmyjazz said:

I very much fear that it will be used to replace actual artists on a massive scale.

Given that it seems creation is far less of an indication of commercial success than promotion. And merit doesn't equal big sales,I feel skeptical.

But I don't know shit.

1 hour ago, TwiceHorn said:

Given that it seems creation is far less of an indication of commercial success than promotion.

Yes, that suuprts what he is saying. You can use it to create for pennies on the dollar

32 minutes ago, G650 said:

Yes, that suuprts what he is saying. You can use it to create for pennies on the dollar

I think the promotion spend will stay the same. The labels will spend pennies on the creation and reap the profits. Real artists will be sidelined.

The only safety net here may be the live performance. Right now, that's where the money is derived. If the fans don't want to give up going to concerts, then there is still an angle in play for the real people.

Edited September 4Sep 4 by jimmyjazz

20 minutes ago, jimmyjazz said:

I think the promotion spend will stay the same. The labels will spend pennies on the creation and reap the profits. Real artists will be sidelined.

The only safety net here may be the live performance. Right now, that's where the money is derived. If the fans don't want to give up going to concerts, then there is still an angle in play for the real people.

Exactly.

I'd love to appropriate Isbell's singing voice and playing, but I don't want anything to do with that goofy looking mofo's likeness.

2 hours ago, jimmyjazz said:

I think the promotion spend will stay the same. The labels will spend pennies on the creation and reap the profits. Real artists will be sidelined.

The only safety net here may be the live performance. Right now, that's where the money is derived. If the fans don't want to give up going to concerts, then there is still an angle in play for the real people.

I’m doing my part. 20-ish shows this year plus I purchase a reasonable amount of vinyl.

I hate the AI slop, but it's sketchy legal ground. Every artist has parts of other artists that inspired them. AI is just doing the same thing,

Most current music was already basically sounding pretty much like AI anyway. Just an average of bits an pieces of what gets clicks and likes, with robotic production and cyber enhanced voicing.

5 hours ago, G650 said:

Yes, that suuprts what he is saying. You can use it to create for pennies on the dollar

What I mean to say is, there are tons of talented performers out there, few of whom "make it big." It's more about promotion than talent or some other appeal.

And, it seems the market would be rather limited for sound/feel/lookalike artists. First accusation of a live one would be "derivative," at best. Imitative or a rip-off artist more likely.

And, for the fans of Isbell and others named here, why on earth would they buy imitative works? I wouldn't buy fake Cracker or CVB. It might be amusing to hear AIs "take" on Take the Skinheads Bowling.

I could see it maybe if you're jonesing for some new Petty or something.

The better the technology, the less obvious it's derivative. Why would a label pay $100K or more to record a new artist's debut album when they could imitate something that harkens back to a known seller for $1K? Slap a name on it, put it out there, the average consumer won't know the difference. Throw some curveballs into the prompts like "Tom Petty backed by The Replacements circa 'Tim' but with Jim Keltner on drums and make the songs about Canada". "Steve Slink and the Waiters". Boom.

Like I said, the problem will be when consumers want to see it live. Ignore the consumers, create a new artist. Lather, rinse, repeat.

Labels in general don't give a shit about art. They want to sell music -- specifically, streams.

EDIT: More correctly "streams, film/TV placement, video game placement, merchandise, etc."

Edited September 5Sep 5 by jimmyjazz

8 hours ago, TwiceHorn said:

What I mean to say is, there are tons of talented performers out there, few of whom "make it big." It's more about promotion than talent or some other appeal.

And, it seems the market would be rather limited for sound/feel/lookalike artists. First accusation of a live one would be "derivative," at best. Imitative or a rip-off artist more likely.

And, for the fans of Isbell and others named here, why on earth would they buy imitative works? I wouldn't buy fake Cracker or CVB. It might be amusing to hear AIs "take" on Take the Skinheads Bowling.

I could see it maybe if you're jonesing for some new Petty or something.

You are looking at this the wrong way by approaching it as a music fan. The vast, vast majority of people consider music window dressing. Hooky noises coming out of a box is the only requirement, plus the majority of music produced is for sync purposes in TV shows, movies, commercials etc anyway. All the artists for that will be out of work. And even the cheapest starving artist costs 100x what an AI generated work does.

This discussion is not really totally about this thread though as whether it's copying a known artists likeness or not isn't affecting the end result beyond just general caricature.

As among the least musical people ever in terms of creative or performing talent, I'm sure that's right.

1 hour ago, G650 said:

And even the cheapest starving artist costs 100x what an AI generated work does.

I wonder, though, as the true cost of AI and its use comes to "bear," how true this will be.

Also, I suppose the "commodity music" to which you refer can be produced easily enough without coming anywhere close to imitating a known artist.

Back on the topic of the sort of legal ins and outs here, one very oddball thing is that it has plaintiffs (it's a class action, so many "similarly situated" plaintiffs), residing in different states.

And NiL is generally a creature of state law, although the federal Trademark Act to an unclear extent includes or subsumes various common law theories of "unfair competition."

So the classes are broken down by states and that state's particular law is applied to those plaintiffs.

This is a summary of Texas' applicable law. https://www.dmlp.org/legal-guide/texas-right-publicity-law

You can use that site to look at what others provide too. It's a pretty good site for getting a quick, pretty accurate thumbnail of a states media-related laws.

On 9/4/2026 at 3:04 PM, jimmyjazz said:

I think the promotion spend will stay the same. The labels will spend pennies on the creation and reap the profits. Real artists will be sidelined.

The only safety net here may be the live performance. Right now, that's where the money is derived. If the fans don't want to give up going to concerts, then there is still an angle in play for the real people.

The primary reason why I’ve pivoted from songwriting/recording to live performance. The big artifact of the shit I’ve composed, recorded and released to streaming platforms is to provide my way of musical thinking as training samples so AI models can incorporate my musical DNA in generating their output.

Two watershed moments to come for live music. 1. When requests at cover band gigs start including hits made by AI. 2. When there are similar advancements in robotics as had been had in AI. Then the live music scene will be on the clock.

4 minutes ago, Goredho said:

When there are similar advancements in robotics as had been had in AI.

Man that ship sailed 40 years ago

I struggle with the idea that people will spend similar amounts of money on "live" music performed by robots, butt I'm usually too optimistic about the American consumer.

1 minute ago, jimmyjazz said:

I struggle with the idea that people will spend similar amounts of money on "live" music performed by robots, butt I'm usually too optimistic about the American consumer.

Yeah, I mean, nothing is a given and there is a significant backlash going on against AI, but should real protections fail to materialize, what’s the live music scene going to look like in another 10, 25 or 50 years when AI and robotics converge to allow both digital and physical representations of an artist that are largely indistinguishable by the senses of the average music consumer? Would you have guessed AI would be doing what it is in music today 15 years ago? Certainly the monied interests will try to lower their costs and increase their profits. Get ready for private equity owned Robo-Prince, live and on tour.

There are children being born today who will grow up AI native. They will not know a world from before when AI had infiltrated the arts. Its usage and consumption will be drilled into them relentlessly. And then there will be the day when the last human who knew of a world before AI dies, and there will be no delineation between the consumption of human and artificially generated creativity and performance

i won’t be around long enough to see that last landmark moment, but for my kids, the AI-less world will be a vague memory if remembered at all — unless something diverts our current trajectory.

Edited September 6Sep 6 by Goredho