“Suno’s training data includes essentially all music files of reasonable quality that are accessible on the open internet.”
“Rather than trying to argue that Suno was not trained on copyrighted songs, the company is instead making a Fair Use argument to say that the law should allow for AI training on copyrighted works without permission or compensation.”
Archived (also bypass paywall): https://archive.ph/ivTGs
Great AI gets more rights than us too!
You’re free to learn from any piece of music too. Whether AI is actually learning is still debatable but you have the same rights right now.
I’m still on the edge tbh I feel like it is learning and it is transformative but it’s just too powerful for our current copyright framework.
Either way, that’ll be such a headache for the transformative work clause of copyright for years to come. Also policing training would be completely unenforcable so any decision here would be rather moot in real world practice either way.
Also policing training would be completely unenforcable
That’s where laws would come in. Obviously it would have civil law, not criminal law, but making sure it would be enforceable would have to be part of such laws. For example, forcing model makers to disclose their training dataset in one way or another.
But you can already train models at home also you can just extend existing models with new training data. Will that be regulated too? How?
We are free to learn, but learning is not free.
Freedom vs cost. One cannot pickup a skill without time, effort and more importantly access to guidance and a vast library of content. Same applies to man or machine. The difference is how corporations have essentially reinvented piracy to facilitate their selfish ends after decades of dictating what’s right with DMCA, DRM and what not.
deleted by creator
Also policing training would be completely unenforcable
But…it’s already been enforced, several times.
I think you’re allowed to listen to every song on the open internet too.
But not making business out of them.
If you get an idea from a song, you are 1000% free to turn that into new art. This is the fair use argument.
I agree with the logic, but I don’t think it should apply to LLMs—a humans-only law, if you will.
I think the key is that an LLM can’t have ideas. Its creative endeavors aren’t creative. Art is about the craft and the message and an LLM lacks that context. Like. The best an LLM can do is produce the kinds of music Drake does that is meant to pacify people into continuous consumption
I had a similar thought after I wrote this. LLMs aren’t creating anything so much as style-copying. They’re unique productions, insomuch as rearranging notes or pixels makes something unique, but I think creativity requires conscious agency, which LLMs definitely do not have.
Also, I don’t need to copy the entirety of Drake’s discography to produce music like his, which is an aspect of human creativity that LLMs currently lack.
You can make a business as soon as you’re done listening to them all.
One has to pay a very high cost to do this. These AI companies did not pay. Why do AI companies get a pass on copyrighted material that the rest of us are getting sued, imprisoned, and fined for accessing?
Lot of people flying in this thread to down vote people saying that these media companies live by a different set of rules than the rest of us without understanding that this AI model is basically a huge automated record scratching DJ that can only regurgitate things its heard before reassembled and presented as new. If any of us tried to do this same thing they’d sue our pants off for piracy and plagiarism. But when they do it it’s fine.
plagiarism isn’t a tort.
But I’m not allowed to remix them. That’s the point that’s being made
It can’t remix either so that’s not an issue
That’s the only thing it does.
no, it doesn’t.
woosh
There’s nothing stopping you from going to youtube, listening to a bunch of hit country songs there, and using that inspiration to write a “hit country song about getting your balls caught in a screen door”. That music was free to access, and your ability to create derivative works is fully protected by copyright law.
So if that’s what the AI is doing, then it would be fully legal if it was a person. The question courts are trying to figure out is if AI should be treated like people when it comes to “learning” and creating derivative works.
I think there are good arguments to both sides of that issue. The big advantage of ruling against AI having those rights is that it means that record labels and other rights holders can get compensation for their content being used. The main disadvantage is that high cost barriers to training material will kill off open-source and small company AI, guaranteeing that generative AI is fully controlled by tech giant companies like Google, Microsoft, and Adobe.
I think the best legal outcome is one that attempts to protect both: companies and individuals below a certain revenue threshold (or other scale metrics) can freely train on the open web, but are required to track what was used for training. As they grow, there will be different tiers where they’re required to start paying for the content their model was trained on. Obviously this solution needs a lot of work before being a viable option, but I think something similar to this is the best way to both have competition in the AI space and make sure people get compensated.
I think the solution is just that anything AI generated should be public domain.
That’s the current status quo.
If you use a tool, let’s say photoshop, to make an image, should it be of public domain?
Even if the user effort here is just the prompt, it’s still a tool used by an user.If you roll a set of dice, do you own the number?
I don’t think it is a tool in the same sense that image editing software is.
But if for example you use a LLM to write an outline for something and you heavily edit it, then that’s transformative, and it’s owned by you.
The raw output isn’t yours, even though the prompt and final edited version are.
If you snap a photo of something, you own the photo (at least in the US).
There’s a solid argument that someone doing complex AI image generation has done way more to create the final product than someone snapping a quick pic with their phone.
One could also say that building a camera from first principles is a lot more work than entering a prompt in DALL-E, but using false equivalents isn’t going up get us very far.
I think a fairer comparison in that case would be the difficulty of building a camera vs the difficulty of building and programming an AI capable computer.
That doesn’t really make sense either way though, no one is building their camera/computer from raw materials and then arguing that gives them better intellectual rights.
Well, the AI doesn’t do all the work, you use public domain material (AI output) to create your own copyright protected product/art/thing etc.
All you have to do is put some human work into the creation. I guess the value of the result still correlates with the amount of human work one puts into a project.
It should be fully legal because it’s still a person doing it. Like Cory Doctrow said in this article:
Break down the steps of training a model and it quickly becomes apparent why it’s technically wrong to call this a copyright infringement. First, the act of making transient copies of works – even billions of works – is unequivocally fair use. Unless you think search engines and the Internet Archive shouldn’t exist, then you should support scraping at scale: https://pluralistic.net/2023/09/17/how-to-think-about-scraping/
Making quantitative observations about works is a longstanding, respected and important tool for criticism, analysis, archiving and new acts of creation. Measuring the steady contraction of the vocabulary in successive Agatha Christie novels turns out to offer a fascinating window into her dementia: https://www.theguardian.com/books/2009/apr/03/agatha-christie-alzheimers-research
The final step in training a model is publishing the conclusions of the quantitative analysis of the temporarily copied documents as software code. Code itself is a form of expressive speech – and that expressivity is key to the fight for privacy, because the fact that code is speech limits how governments can censor software: https://www.eff.org/deeplinks/2015/04/remembering-case-established-code-speech/
That’s all these models are, someone’s analysis of the training data in relation to each other, not the data itself. I feel like this is where most people get tripped up. Understanding how these things work makes it all obvious.
I think AI should be allowed ti use any available data, but it has to be made freely available e.g. by making it downloadable on huggingface
Inspiration ≠ mathematically derived similarity.
These aren’t artists giving their own rendering, these are venture capitalists using shiny tools to steal other people hard work.
4 chords.
Yea, I think only big media corporations would profit from such a copyright rule Average Joe’s creations will be scraped because he has no funding to prove and sue those big AI corporations
Taking other people’s creative works to create your own for-profit product is illegal in every way except when AI does it. AI is not a person watching videos. AI is a product using others’ content as its bricks and mortar. Thousands of hours of work on a project you completed being used by someone else to turn a profit, maybe even used in some way you vehemently disagree with, without giving you a dime is unethical and needs regulation from that perspective.
That’s covered by section 107 of the US copyright law, and is actually fine and protected as free use in most cases. As long as the work isn’t a direct copy and instead changes the result to be something different.
All parody type music is protected in this way, whether it’s new lyrics to a song, or even something less “creative” like performing the lyrics of song A to the melody and style of song B.
Taking other people’s creative works to create your own for-profit product is illegal in every way except when AI does it.
No, actually its completely legal to consume content that was uploaded to the internet and then use it as inspiration to create your own works.
Algorithms don’t have “inspiration”.
Taking other people’s creative works to create your own productive work is allowed if you are making a fair use. There’s a very good argument that use such as training a model on a work would be a fair use under the current test; being a transformative use, that replicates practically no actual part of the original piece in the finished work, that (arguably) does not serve as a replacement for that specific piece in the market.
Fair use is the cornerstone of remix art, of fan art, of huge swathes of musical genres. What we are witnessing is the birth of a new technique based on remixing and unfortunately this time around people are convinced that fighting on the side of big copyright is somehow the good thing for artists.
Thousands of hours of work on a project you completed being used by someone else to turn a profit, maybe even used in some way you vehemently disagree with, without giving you a dime is
exactly how human culture progresses, and trying to stop it
is unethical and needs regulation from that perspective.
Taking other people’s creative works to create your own for-profit product is illegal in every way except when AI does it.
wrong.
You’re literally on a piracy server. You know about the laws and how hard the corpos crack down on us. Why the fuck are you licking the boots now
i’m principled, so i know that copying content is good and stopping people from copying is bad.
you’re wrong on the facts, this has nothing to do with supporting corporations.
Sure, jan
Actually, that music was based off of getting royalties and ad viewership. No one will pay for an ai to be exposed to an ad or pay royalties for an ai to hear a song. Or have an ai to hear a song for the chance of the ai buying merchandise or a concert ticket.
Say what you will, but “I Glued My Balls to My Butthole (Again)” fucking slaps.
Sadly I love that song. It resonates with my lived experience.
Link please? For research purposes, of course.
https://youtu.be/wPlOYPGMRws?si=Xu094G4Ta5HfeLKS
YouTube link
https://www.youtube.com/watch?v=wPlOYPGMRws&list=OLAK5uy_n2EyGSzgPcGJtF3-inSboC-aU3Nm8jKGE
Listen to the rest of them too. This guy is a riot. He writes the lyrics and records parts of the music to make it better than the AI can make it.
I prefer ‘I Stink’.
Why would the “AI” know about it and not just make that up though? But also, if companies can use everything on the internet as a “fair use” thing, then so should people. No more bullying of people making stuff you don’t like / agree with.
If copyright had a reasonable duration, a huge chunk of that would have been public domain and not an issue.
If you post stuff online for people to view it, don’t be surprised when computers see it too.
Posting doesn’t mean anyone can exploit it for commercial gain. Also a lot of music is pirated against the Creator’s wishes.
Copyright doesn’t work like that.
The fact that you put online (e.g. in an online shop or, heck, even for free on your website) doesn’t imply anyone can use it for anything they please.
For example an mp3 on a indie musician website for making people know their music, doesn’t mean people can start making CD out of it and selling them
You may say that piracy exists but it is illegal and AI training is pretty much for profit piracy (using something outside the intended scope defined by the author for profit)
I think that the argument is that it’s not actually copying it and instead learning from it.
I feel like they really have to pay something. Maybe their argument is right and they shouldn’t pay full price given that if a human did the same thing to learn to make music, they also wouldn’t be billed in that way, given it’s openly available.
Maybe they are required to give a portion of profits back to the source materials creators?
The purpose of the copyright system is that someone should not be allowed to take your material and sell it to someone else passing it off as your own and make the profit thereby stealing the profit from you. I don’t think that’s what’s happening here but also at the same time it doesn’t seem like what they’re doing is right because they’re making profit off of somebody else’s work without paying anything.
It feels to me like the answer should be somewhere in the middle.
I certainly don’t have the answers though.
Sure, this company will burn for this, but Pandora’s box is wide open now.
I’m not condoning anything, but the original comment is unfortunately 100% true.
Yea, it’s the victim’s fault!
AI LLM SVU over here
UMG can go to hell. Musicians already make almost next to nothing from song sales because the labels take so much.
Yeah but the fucking labels aren’t even being that resistant to AI. AI is robbing the labels a little, and the artists a lot. It’s like how in 1984 george orwell points out the proletariat and the middle class should be allies, but the middle class always installs itself as the upper class and leaves the proletariat behind. The labels are willing to ditch the artists so long as they get to stay in one of the upper two classes. So they’re willing to sell access to their back catalog and their lost masters to the AI companies so long as that pays them a little bit better than paying out for new artists to join the label. They’re the fucking overseers or the small whites in Haiti. They hurt the people they should be working with because they’re too short sighted to see what’s gonna happen once everything shakes out. And you’re over here rooting the AI robbery on because it will hurt the labels instead of realizing who it’s REALLY gonna hurt
Abolish copyright.
I’m fine with copyright, provided it’s limited to only a few years and can’t ever be extended. This “lifetime of the author plus 50 years” shit is what makes it terrible.
deleted by creator
I think you have to abolish hierarchical society first. Which I also think should be done. But if you get rid of rules protecting creative endeavors before getting rid of the shitheel corporations it gets rid of creative endeavor more than it gets rid of the corporate bastards. The problem with copyright as it stands is how much the corporate bastards have twisted it over time to benefit them instead of the collective us.
You don’t have to abolish hierarchy, you just have to de-tangle work from sustenance-level resource distribution. A UBI sufficient for living would be enough. Even providing universal housing and reducing the workweek would help.
Copyright simply makes creative work profitable, but profit isn’t necessarily a prerequisite for creative work.
Oh my desire to abolish hierarchical society is… Pretty core to who I am lol. So there was some bias in that. Yeah there’s ways to make get rid of copyright, still have professions, and still have a hierarchy, but at a certain point you’re building a gentler form of capitalism instead of treating people with the true respect and dignity they deserve. I’m willing to accept I’m pretty radical on this particular set of views.
Also yes. Everyone should get UBI and the billionaires pockets should be where we get the money from
rules protecting creative endeavors
that’s not what copyright does.
Why?
Because it manufactures scarcity and causes us to repeatedly expend energy reproducing things that could be otherwise copied and enjoyed at near-zero cost.
We keep inventing silly rules in order to put off dealing with the existential threat our mode of production represents. “Copyright” is the first and silliest of those rules.
Are you taking about patents? Cause a world without copyright doesn’t sound very fun to me. Or anyone in a remotely creative job.
Ever for patents: There’s a reason innovations are protected literally anywhere in the world, but the durations being ever longer is a real problem (5 years would probably be fine). The basic concept is still just straight up necessary.
No, i’m talking about all intellectual works (copyrights and patents being some of the categories commonly used)
Humans need no incentive to create new works, but the way we distribute resources requires us to make these rules so that those creators fit within our ‘work for food’ production model.
Even a modest UBI would support most creative endeavors, but instead of that we have a “monetize your work or starve” arrangement.
Oh, so when a ai company wants to use copyrighted works without permission or compensation it’s “fair use” but when I do it, it’s “piracy” and “I need to leave before the cops are called”.
Imagine how many years of prison would get an individual if he admitted in court to pirate tens of millions of music files in order to make a profit
I mean Suno is being sued by the record labels right now for this. We will see how much they have to pay if they lose
I don’t know if corpos will allow such dangerous precedent since all mega corps are guilty of this
One of the four fair use factors is the portion of the copyrighted work that was taken. For a finding of fair use under this factor, the infringing work must only take the amount of copyrighted material needed for the infringing work’s purpose.
If they ripped every single file they have access to, there’s no way to be found as fair use under this factor. If they argue they were using a curated list of only the works they needed to develop their model it could be fair use, but admitting to taking every possible work in their entirety is a surefire way to fail a fair use defense.
Your interpretation of copyright law would be helped by reading this piece from an EFF lawyer who has actually litigated copyright cases in the past:
https://www.eff.org/deeplinks/2023/04/how-we-think-about-copyright-and-ai-art-0
But that’s not how model training works, it doesn’t simply copy and paste entire songs into its training data. It more or less “listens” to it, analyzes it and when you ask to create a rock song for example it just has an algorithm behind it what a song like that would sound like.
But you can’t just ask it to generate Bohemian Rhapsody from its data, it would probably get very close depending on the training, but it would never be 100% the same (except the model was only trained on this one song).
Just like you can listen to rock songs and then make your own, that’s totally valid. The problem here is of course automation and scale, but saying it’s not fair use is dubious.
Fair use is a legal doctrine relating to derivitave works based on copyrighted works. An AI model’s fair use determination would be judged by the same standards and all derivative works.
It doesn’t matter how they used the copyrighted works. This factor is about scale not intent.
There are four factors, and no single factor is determinative. But admitting their model uses as much training as possible makes their model less likely to be fair use.
If I as a human listened to every single song of a band from start to finish, then produced a similar song in the same vein (lyrics / music genre), it would be fair use.
So why would it stop being fair use if an AI does the same thing? Just that the AI can listen to every song of this band and a million other bands, combining them.
Because fair use is an affirmative defense to copyright infringement. To use a fair use defense you have to admit your work is infringing, but argue that the infringement is justifiable.
Trying to defend the AI with fair use requires you to admit the AI itself is infringing, but justifiable, and by the doctrine of fair use, it is almost certainly not.
Only humans can hold copyrights. Your example would be a non-infringing work because it lacks direct copying. An AI doing the same would make an uncopyrightable work, with the AI itself being infinging if you tried the fair use defense.
Only humans can hold copyrights.
Yeah, no. Most copyrighted material is owned by companies, you don’t have to be a natural person to hold copyrights. And if a company can hold copyrights, you can also argue it can have fair use.
Companies are run by people. The human employees create copyrighted works that become the property of their employer by the terms of their contract. That’s how work for hire contracts work…
You would know this if you have ever worked in any creative field.
why are comments full of AI shilling? are these bot accounts or are there still real people actually defending “AI” ““art””?
Yeah, imagine different people having different opinions than you!
Can’t be, must be russians.
on “AI art” being art, no, i don’t believe an actual human being can believe a fucking algorithm that can’t tell what itself is doing is capable of artistic expression.
Because regardless of whatever data Suno trained their model on, the stuff it makes sounds like shit and won’t replace real music, just like Midjourney isn’t replacing real art.
they’re right, they should absolutely be allowed to use and profit off of other peoples work that they spent decades learning and perfecting
(this is sarcasm in case you can’t tell they’re dumb as fuck for saying that shut this company down today pls)
It’s pretty obvious if you get specific with the tags. Especially with older styles where there’s less available training data.
Edit: just go to the album’s page on RYM and copy the tags to try for yourself
Just tried it out out of curiosity. It does a good job, but the music is still boring.
Imagine the amount of data that is used to train facial recognition ia.