These tired arguments for AI art are irrelevant and all over the place. Nobody cares about the model output or an academic explanation of how diffusion works. It is all about the training data.
The issue at hand is that artists had their art trained on without their permission and violating the copyright.
I've just searched the dataset to find mickey mouse, marvel and some disney characters images in the training dataset.
So then it is trained on copyrighted material, and Midjourney and especially SD are building and profit SaaS on the backs of the work without permission or compensation.
Even that alone is a valid reason litigation, and I wouldn't be surprised if Disney joins in against this or asks SD to remove them from the next model release.
Just do the following:
1. Just ask for permission to use the artists work?
2. Train on public domain images, there are plenty out there.
> artists had their art trained on without their permission and violating the copyright
Okay, I'll bite. Since when does copyright require permission to learn from the copyrighted material? I don't need your permission for that, and neither should an AI.
Or alternately, assuming you reject the "human-AI analogy": since when do I need permission to collect statistical information about copyrighted material? Should Disney be able to sue me for telling you how much screen time Jafar had in Disney's Aladdin, or what percentage of pixels in the first 5 minutes were a shade of blue?
Nobody's arguing that near-duplications or trademark violations are not bad. But I think they want a broader power grab - to extend copyright to art styles. In other words to protect not just the expression but also the ideas themselves. They want something like patent + trademark + copyright all in one.
Copyright requires permission to produce derivative works.
If I take a picture of Mickey Mouse, add 1 to each RGB value, it would still be covered under copyright.
I think the term 'learn' is doing the heavy lifting in your argument here. Specifically, your argument seems to be:
1. We use the term 'learn' as shorthand for 'fitting a machine learning algorithm to data'.
2. We also use the term 'learn' for a human looking at copyrighted material to practice their craft.
3. Humans using copyrighted material to learn doesn't make their works derivative.
4. Therefore, machine learning algorithms that are trained on copyrighted material aren't producing derivative works.
This argument holds an implicit assumption: if we use the same term for a process, it should covered by the same rules/laws. Which I think isn't very sound.
Can you be more specific? I'm not sure how this affects the argument I've laid out.
To be clear, my argument doesn't show that statistical generative approaches to art are violating copyright by using copyrighted data. It only shows that the argument "humans learn on copyrighted stuff therefore it's legal for AI to do it" isn't a sound argument.
Another argument that relies on the definition of derivative work might be able to show that you can train a model on copyrighted data without producing works that violate copyright.
"humans learn on copyrighted stuff therefore it's legal for AI to do it" - Learning can never be a violation of copyright. A work (digital or physical) can potentially violate copyright.
So there are two potential "work" here:
1. The model itself
2. Images created with the model.
I would argue that (1) can't violate copyright as it's not a derivative work
(2) may or may not violate copyright but if it does, then it does so to exactly the same degree as it would if it wasn't AI generated.
In other words "You're not allowed to plagarize copyrighted works no matter what production method you use".
The model is a copyright violation as it contains untransformed copyrighted material. You can show this by simply asking for a picture of Mickey Mouse.
To preempt a few (obvious) arguments about what it means to be "transformed", keep in mind an encrypted zip file containing images of Mickey Mouse would not be considered transformative, even if it happens to contain images that aren't copyrighted.
(2)
> then it does so to exactly the same degree as it would if it wasn't AI generated.
This is the conclusion of an argument, but I'd need to see how you show that "it does so to exactly the same degree". Right now, it looks like you're taking that as a base assumption.
> (1) The model is a copyright violation as it contains untransformed copyrighted material. You can show this by simply asking for a picture of Mickey Mouse.
This is a strong argument and gave me pause. I think if Disney sued they could probably win.
However - examples such as this are probably quite rare. Mickey Mouse is so uniquitous that is massively overrepresented in the training and data and that has resulted in overfitting. For the vast majority of artists and almost certainly for the plaintiffs in this case, the demonstration wouldn't work. You most likely cannot produce an infringing image from a concise text prompt.
Arguably one of the outcomes of legal action might be the need for a lot of effort to remove examples similar to the Mickey Mouse example. This would be a good thing but wouldn't prevent AI image generation from broadly doing the things that the plaintiffs want to stop it doing.
> (2) but I'd need to see how you show that "it does so to exactly the same degree".
Not sure I follow? If I produce an infringing image and a court agrees it's infringing then I'm liable. It doesn't matter how the image was produced. I don't see what there is to dispute here?
For (1), I think it's handy to think of an encrypted zip file that contains a bunch of images. If you give it a valid password, it'll give you a corresponding image. If it produces an image that violates copyright, it clearly violates copyright, even if most of the images contained in it don't.
It's possible that if the model required sufficient instruction to reproduce Mickey Mouse ("create a line from (0,20) to (20,30) of thickness 2, then a line from ..."), and could not do so without that, then the argument wouldn't apply. And there's definitely a grey area once we're outside of the territory of reproducing images. But we definitely aren't anywhere close to that with these models, and looking at the "AI Safety" work, it doesn't seem possible to get these models to "behave".
> This would be a good thing but wouldn't prevent AI image generation from broadly doing the things that the plaintiffs want to stop it doing.
Yeah, I agree that the plaintiffs aren't suing solely because the model can easily regenerate their images. But I think we'll find its insanely hard to get these models to avoid producing copyrighted images. They may be forced to license the training images.
You directed the model to create an image of Mickey Mouse. You would also be the one who would then decide to 1.) hang that picture on your wall which is totally legal even if you photocopied a still from a Disney cartoon or 2.) put that image on a t-shirt and sell it.
I don't think this is really a legal interpretation.
If someone was distributing a password protected zip-file where if I type in "Mickey Mouse", I get a photo of Mickey Mouse, the zip file violates copyright. It doesn't matter that the underlying representation of Mickey Mouse is distorted in storage and recovered later. It doesn't matter that in order to retrieve the image, I have to go through a series of operations.
Of course, if I opened the zip file and put the image on a t-shirt, I would also be violating copyright. But the zip file itself is copyright infringing.
If you print out the 1s and 0s that make up the zip file and sell that on a t-shirt then you are not infringing on a copyright. If you buy that shirt, scan the 1s and 0s, unzip, print and sell the resulting image on a t-shirt, the you are infringing.
At this point we can just wait until the courts clarify this for you because you seem to be missing a fundamental aspect of how our legal system operates. I’d much rather go and watch football.
However, what stable diffusion does bears no resemblance to "add 1 to each RGB value" and bears substantial resemblance to the human process of "learning".
Without litigating if the model is "learning" in a human analogous way, instead I think the burden is on showing that the model is "substantially copying" and to an equal or greater degree than what would be needed to consider a work derivative.
Another counterargument is that these models do not generate anything without a prompt. So in a way it’s like a very smart photoshop: if you, a designer, use tools to create a character that resembles Mickey Mouse - this is on you, not the tool. It’s going to be interesting to hear this being argued in court.
A password-protected zip file containing Mickey Mouse photos also doesn't produce copyrighted material without the right prompt. Yet, the distributor of the zip file is definitely responsible for copyright violation.
And to preempt a likely counterargument, intent is not a required factor for establishing copyright violation.
The important distinction is that the model does not contain a copy of a copyrighted material. It contains data trained on copyrighted material (like our brains).
You're making a distinction that isn't reflected in copyright law.
If I take a JPEG of Mickey Mouse and then turn it into a PNG, it's not a "copy", as the bits are different. But it still contains copyrighted material.
You can try to argue that the bits of PNG itself aren't an image of Mickey Mouse, but rather the algorithm that reads the PNG produces an image of Mickey Mouse. But that isn't really a relevant distinction in so far as copyright is concerned.
In addition, this statement is false:
> The important distinction is that the model does not contain a copy of a copyrighted material.
It has been shown repeatedly that the model produces copies of training data. The copies are of course not stored as JPEGs/PNGs in the model, but they are retrievable from the model, given the correct password (prompt).
Could you provide evidence of your last statement? I haven’t seen these models produce actual copies of any art (can’t imagine that’s an option in general).
These models do not contain copies. One way to describe the data is they contain a statistical breakdown of the artwork, which is substantially different from a JPEG -> PNG conversion you mention.
> Could you provide evidence of your last statement? I haven’t seen these models produce actual copies of any art (can’t imagine that’s an option in general).
> These models do not contain copies. One way to describe the data is they contain a statistical breakdown of the artwork, which is substantially different from a JPEG -> PNG conversion you mention.
I don't understand the distinction you're making. What legally separates a "statistical breakdown" representation from a zip file representation, JPEG representation, PNG representation?
As I anticipated, you are referencing research that does not show exact copies being generated by Stable Diffusion. Do "semantically equivalent" images infringe on copyright? I would argue that they do not. We will see how this plays out in court.
Food for thought: if I write instructions for generating an SVG of a black square, does my program contain copyrighted material (Malevich's Black Square)? You and I could argue about that, but you will probably quote more research that disproves your own point. So let's skip that.
> As I anticipated, you are referencing research that does not show exact copies being generated by Stable Diffusion. Do "semantically equivalent" images infringe on copyright? I would argue that they do not. We will see how this plays out in court.
If you're convinced that a photo of Mickey Mouse with slightly larger ears, or slightly reddish pants isn't copyright infringement, then sure, neither is any of this stuff. It would also mean that republishing copyrighted images with lossy compression algorithm (IE, JPEGs) would also not violate copyright.
I would suggest looking at the actual laws around copyright instead of relying on what you feel copyright should be.
Amazing. It’s so obvious I wonder why billion dollar corporations didn’t figure out the legal implications of these models yet. Do you have an email address I could pass on to OpenAI?
> If you're convinced that a photo of Mickey Mouse with slightly larger ears, or slightly reddish pants isn't copyright infringement, then sure, neither is any of this stuff.
Isn't this already well-established? For example, this image, used in The Simpsons:
To be clear, I'm not saying that you can't create Mickey Mouse images that are transformative (or that Disney might not bother suing over; I think there'd be a lawsuit if the Simpsons tried making a commercial film following the adventures of their rendition of Mickey Mouse).
Also, that usage of Mickey Mouse might be copyright infringing, but fall under fair use (probably parody), which is a specific defense of copyright infringement (similar to "self defense").
What I am saying is that:
1) If your model returns images which look near-identical to your training data, then any copyright infringement that applies to the training image will also apply to your image.
2) If your model can consistently return copyrighted imagery, there's little difference between explicitly sharing those images (with a password) and implicitly sharing them (via a model + prompt).
> intent is not a required factor for establishing copyright violation.
What you mean to say is that if someone just so happens to write a story that is significantly similar to another work that their intent doesn't matter... most importantly their intent is unknowable.
This doesn't mean that courts will not be interested in the intended use of specific mechanical tools such as VCRs, computers, and statistical models.
Let's tease apart this "zip file containing Mickey Mouse photos" because it seems to be from where your errors in reasoning stem.
Here is a website that gives you instructions on how to draw Mickey Mouse:
These instructions are not violation of copyright because they draw a picture of Mickey Mouse. Copyright does not protect the mere existence of an image. Copyright gives a monopoly on distributing copies of that image to the rights holder. Not the idea of the image, not the instructions that could lead to the image, not the 1s and 0s that make up the image.
As I've mentioned before, a t-shirt with the 1s and 0s that make up the zip file that contains those photos is not in violation of copyrights on those images. You'll notice that CafePress still has t-shirts with the DeCSS code for sale: https://www.cafepress.com/+,954530397
Imagine a mathematical formula that when plotted in polar coordinates draws Mickey Mouse. The mathematical formula is a fact and it cannot be copyrighted. There's another step required for copyright infringement: Someone renders the image and distributes copies of the rendered image.
I mentioned VCRs specifically because in 1976 Universal Studios sued Sony over the claims that because Sony manufactured a device that could be used for copyright infringement the company was liable for infringement committed by consumers of Betamax.
Take a note of the majority opinion in the Supreme Court ruling:
> The question is thus whether the Betamax is capable of commercially significant noninfringing uses ... one potential use of the Betamax plainly satisfies this standard, however it is understood: private, noncommercial time-shifting in the home.[7] [...] [W]hen one considers the nature of a televised copyrighted audiovisual work... and that time-shifting merely enables a viewer to see such a work which he had been invited to witness in its entirety free of charge, the fact... that the entire work is reproduced... does not have its ordinary effect of militating against a finding of fair use.
Stable Diffusion is also capable of commercially significant non-infringing uses. It is to be expected that the lower courts will continue to follow the precedence set by the Supreme Court that there is a clear distinction between a mechanical device capable of making copies and the distribution of those copies.
To put in the context of Stable Diffusion, this is why legal experts keep saying there are two things to consider: The model, which is almost guaranteed to be protected by fair use, and the outputs of the model, in which the users of the model are liable for any copyright infringement.
1) This is covered by a trade secrets law, and not copyright.
2) It still being for sale doesn't mean it doesn't violate copyright; only that the copyright holder hasn't litigated to stop it. Put differently, getting away with something doesn't make it legal.
Look, I think Stable Diffusion is pretty neat. But all of this "it's just math"/"it's just data"/"it also creates non-copyrighted images" stuff doesn't cut it as far as the law is concerned. If your point is that being in possession of the model isn't violating copyright, sure.
I'm talking about distributing or otherwise providing a service that uses the model to generate images. If they figure out how to get it to stop generating training data, they'll be fine. But with existing approaches, that'll be nigh impossible.
> Now this is plainly false. I cannot start selling a Mickey Mouse JPEG stripped straight from Disney's site.
You can absolutely go to the Disney website right now, download any JPEG, convert that image to a series of 1s and 0s, and then print that series of 1s and 0s on a t-shirt that you sell on your website.
Once those 1s and 0s are in a new non-digital medium they are no longer a copy of the image. Historically, the digital copy is a special case because practically there is no difference between the "expression" of a byte sequence in a digital medium and the "idea" of a byte sequence in a digital medium.
A t-shirt is not a digital medium so the rights holders don’t get to extend their monopolization of speech to a screen print of an unintelligible pattern of 1s and 0s.
> A t-shirt is not a digital medium so the rights holders don’t get to extend their monopolization of speech to a screen print of an unintelligible pattern of 1s and 0s.
Can you point me to a ruling where this form of expression is successfully defended via fair use?
Seems unlikely that a judge would agree that a simple transformation between "digital" vs "non-digital" medium avoids copyright infringement (ex: if I produce a digital image, I definitely maintain copyright over digital and non-digital forms of that image; you can't just start printing and selling posters of my image).
You could try to argue that the change of medium is transformative, but that's not a generic argument for "digital" vs. "non-digital"; that's particular to the circumstances of the case.
What are you missing about the fact that a t-shirt with Mickey Mouse on it and a t-shirt with a pattern of 1s and 0s are completely different. Is anyone buying and wearing the 1s and 0s shirt instead of buying and wearing a Mickey shirt?
You can’t say the same thing about digital copies! Is anyone buying the 1s and 0s of a JPEG in a digital medium and expecting to just have a bunch of 1s and 0s?
BTW, if anyone wants access to those 1s and 0s they only need to go to the Disney website to download the JPEG… they don’t need to buy my 1s and 0s shirt.
You’re just not thinking about this in legal terms: who has wronged who and how did they wrong them. These lawsuits are not criminal, not public wrongs, but rather private wrongs. People must show that there are concrete damages against their specific property!
> But all of this "it's just math"/"it's just data"/"it also creates non-copyrighted images" stuff doesn't cut it as far as the law is concerned.
What you mean is that you would prefer if Stable Diffusion was not considered fair-use by our legal system.
However, based on my more than cursory understanding of copyright law in the US, it seems more likely that the courts will find these tools fair-use than otherwise.
> What you mean is that you would prefer if Stable Diffusion was not considered fair-use by our legal system.
I think you misunderstand me. Despite my feelings about Stable Diffusion and laws around copyright, I believe the law will find in favour of the artists, assuming they have the funds/will to see it through. The law unfortunately doesn't really care how I feel about it.
Granted, they're up against some fairly large corporations that have both time and money to stall for a while, so I don't know how far their lawsuit will go.
> If they figure out how to get it to stop generating training data, they'll be fine.
How does that change anything at all? Take any of the plaintiffs in the Stable Diffusion lawsuit. Let’s say that their work was left out of the training data or if it was allowed to remain that the tool never reproduced training data… is Stable Diffusion any more or any less useful? Is literally anyone using Stable Diffusion with the intent to reproduce one of the plaintiff's copyright protected works?
> Let’s say that their work was left out of the training data or if it was allowed to remain that the tool never reproduced training data… is Stable Diffusion any more or any less useful?
This doesn't really have much to do with copyright.
> Is literally anyone using Stable Diffusion with the intent to reproduce one of the plaintiff's copyright protected works?
Intent isn't really a factor in terms of damages for copyright infringement. But I assume we'll figure out the extend of the damages if/when damages are awarded, since that's when they measure it out. Granted, I expect the liable parties to settle far before it gets to that; the payout seems cheap (~100k - 1m) relative to the risk (penalties for copyright infringement are per unit, and it's easy to hit millions of units on the internet).
And yes, a lawyer for the defense will show that, unlike a zip file of photos that can ONLY be copies of protected works and that anyone distributing such a zip file cannot reasonably say that they are expecting non-infringing works, with Stable Diffusion there is basically no one expecting to get copies of any of the plaintiffs protected works.
Like, literally no one who made or uses SD knows or cares about the plaintiffs, is interested in their drawing style, and even remotely interested in any of their individual protected works. If I were the defendants I would train a model without any of the plaintiffs’ work and then show the court that the results for an astronaut riding a horse are no worse without their works.
And no, some lawyer is not going to scream out “objection, intent has nothing to do with copyright”, because, again, all that is meant by the intent is this: if Stable Diffusion output an image but your intent was not to copy another image… well too bad, you’re still infringing… this all happens where the liability for disseminating the copy and foregoing any fair use is on the person using the tool. Is using the tool worth the risk? Maybe not!
The intentions of the toolmakers are very important to the courts. A tool like a torrent website that does literally nothing to police their content and even actively promotes piracy is going to be treated differently than publicdomaintorrents.info.
> > Let’s say that their work was left out of the training data or if it was allowed to remain that the tool never reproduced training data… is Stable Diffusion any more or any less useful?
> This doesn't really have much to do with copyright.
It is not specific to copyright but any private wrong heard by the courts needs a clearly defined plaintiff, defendant and property in question.
How can the plaintiffs accuse anyone of infringing on their copyrights when no one has used the tool to produce the protected works? Which of their specific works was infringed upon? Is that specific work a requirement for the tool to function?
The plaintiffs seem to be “everyone” and their property “every single digital image”, which is legally incoherent. How exactly would someone reasonably ask permission for every single digital image? Is everyone entitled to part of the settlement or just the plaintiff’s and their lawyers?
> And yet torrent sites have been found to violate copyright, despite not hosting any of the copyrighted content.
BitTorrent itself is legal. The torrent sites that were shut down were found to have basically no non-infringing utility. There were and still are other torrent sites that have non-infringing utility which are of course allowed to exist.
Yes, and so is Winzip, and the code that trains Stable Diffusion. But that doesn't have much to do with the resulting data that is used and generated by those files.
> There were and still are other torrent sites that have non-infringing utility which are of course allowed to exist.
Can you point me to a torrent site that withstood a legal challenge for copyright infringement?
Exactly. The model only produces copyright protected material when the operator of the model directs the sampler towards the part of the latent space that contains that material.
The user is liable. The user is also liable for using a photocopier to create and sell copies of someone else’s work. Anyone can photocopy anything they want and plaster their own walls with that work. In all cases the manufacturer of the photocopier is not liable…
Now, if these tools consistently plagued users with being liable for copyright violations then these tools become more trouble than they are worth.
> Copyright requires permission to produce derivative works.
This then would put the responsibility on the creator, not the software. If I use a prompt "Mickey Mouse fighting Thanos" then the images produced are clearly derivative. But is this fair use? There's plenty of fan art out there like that. In fact, I typed that string into google and got fan artwork of Mickey as Thanos and the infinity stones are different media centers that Disney controls[0]. Making the image quite a high quality parity. I'm not a lawyer, but I don't feel like this image should be illegal. If so, doesn't that make this conversation far more complex than what you're writing here? I'd agree if I was trying to sell pictures of Mickey Mouse and claim to be Disney (and that there was no case for parody), but this does seem like a really hard line to draw and I don't think people are being honest about that.
Exactly, the person coaxing copyrighted material from the latent space is liable.
Now, a tool that burdens an individual with a torrent of copy-written material might not be worth using, but that is separate from the data embedded in the latent space of the model itself.
> the person coaxing copyrighted material from the latent space
I'd be careful with this terminology. These models are able to generate more diversity in data than they are able to memorize. So I just want to make sure readers do not read this as "copying and combining things to make an image" in a copy/paste manner rather than the same manner that humans do (very different mechanisms) of having a broader understanding of shapes and fuzzy concepts. But I'm also a generative modeling researcher and so I'm biased to be pedantic.
You definitely need permission to make copies of people's art on your computers to use whilst learning though (copyright holders tend to be OK with or completely unaware of that, but it's a right they reserve)
More to the point, this feels like the "have your cake and eat it" of legal arguments. If we're going to anthropomorphise AI to the extent we argue storing data in its data corpus without authorisation is just like allowing a human to look at your legally owned stuff, and reinterpreting that data in a [sometimes not particularly] lossy fashion as a set of weights and biases is just like a human learning, we should probably follow the "AI is just like a person" argument through to its logical conclusion and hold it legally responsible for not respecting copyright law if it ever produces outputs with trademarked images or uncanny resemblance to copyrighted material that aren't deemed "fair use", just like a person would be.
(tbf, it's not too hard to imagine a world in which ChatGPT says "I'm sorry, but I am not able to write stories involving Mickey Mouse, because Mickey Mouse is the intellectual property of the Disney Corporation. Respecting intellectual property rights is very important", and I'm not sure that's an improvement...)
> Since when does copyright require permission to learn from the copyrighted material?
Ah, but you are anthropomorphizing an algorithm. What is happening inside the computer you call learning, but it is just some mathematical function. Other people might not agree that this is learning, and might call it "mixing" or "copying".
Fine then. Say I went through Star Wars and counted how many blue pixels there were in each frame and published a list of the counts. Would you claim that list would be infringing?
Doing statistical analysis does not violate copyright, or Google would have to shut down tomorrow.
> Fine then. Say I went through Star Wars and counted how many blue pixels there were in each frame and published a list of the counts. Would you claim that list would be infringing?
If you were instructed to counted blue pixels by somebody who used your lists for a SaaS product which could generate high fidelity representations of Death Stars, wookies and Harrison Ford (someone else probably counted the other colour pixels!) every time a customer of theirs typed in "Star Wars", quite possibly.
The algorithm itself tries to learn to remove noise. This part of algorithm is undeniably a learning process, evidenced by the fact that it gets better along the training process. It just happens to be that, in order to remove noise, you need to have an understanding of the underlying thing(probability distribution), so in order to learn to remove noise, it needs to learn the images. I don't understand how a mathematical function by definition can't learn, since it's just a language we use to describe what's going on.
Copyrighted training data is used in the training of a model, but doesn’t exist within the model weights in any real sense. Doesn’t seem different to if I hired an artist to make me a painting of a Disney character (I’m sure there exists copyrighted material in their training dataset).
If it generates an artists signature when producing something in that artists style, then I’d argue against that idea. And frequently enough, they will generate a recognizable signature.
This is just because it’s still a little dumb. If I hired a human artist to produce something in another other artists style and it copied their signature, what is that considered to be? Artist stupidity imo. Or at least a failure of communication.
A machine is copying the image when learning from it no more than a brain is copying an image by trying to memorize it. In either case you are never going to get the same image out. I am not aware of copyright asserting itself over the perception of anything.
Plus, regardless of poorly-enforced legal 'realities', you cannot control something once it is out in the public space. It may as well be public domain. Artists are learning first-hand that they have no control, as it should be.
I say this as an artist myself, either I release my work publicly and it will do with it what it wishes (paying me would be nice, but really difficult to reasonably expect), or I hoard my art privately and it dies when I die, unseen and unperceived.
The problem is the computer people have been dealing with bots for a while, we have tools and years of best practice to deal with it. Artists don't, this is new to them, and frankly a lot of them like to get wrapped up in emotions and do not give a damn about technical nuances -- as you correctly assert. So they will generally repeat whatever trite line sounds best. (You can see this firsthand on with the copypasted memes on facebook/reddit)
How many of these guys truly understand the issue? Or are they just the next generation of luddites trying to fight off the next printing press or camera?
No, you can't require fees to be paid. That is my point.
You can try -- all of entertainment certainly does with DRM and all these stupids rental models for everything. They can come up with registries and agencies to handle "collections", but all of this is toothless if I can acquire the bits and do whatever I want outside of their purview.
If I can find a way to view the artwork I can do what I wish with the knowledge and emotions that I gained from it, and there is nothing you can do to stop me. This is not so different from AI learning, the exact images are not stored in the model. Neither are they "compressed". You cannot get the same exact image out that you put in.
People need to get over this awful model of rent-seeking and collections. There are always other ways to monetize.
that's a personal opinion, not a legal fact. until we don't have court rulings on this matter, all legal statements are opinions. in order to have a civilized debate, we should be careful about separating opinions and facts, and clearly marking the former as such.
> The issue at hand is that artists had their art trained on without their permission
Interesting that you care about the training data but not the generated images. The only explanation I found is that you want to stop the "evil" from the root and have the models unaware of your style.
Do you believe that the style of an artist who is not included in the diffusion training set can't be replicated? No style is actually outside the scope of the model after it has learned the first few thousand styles. They become a map that will reveal every other style.
Your argument would make Google illegal -- they are copying and indexing other people's content (including images, as it happens), to make a profit, on the backs of other people's work, without a license or permission.
But the courts have already ruled that this is not the case. Your argument is based on what makes sense to you, but not seemingly informed by actual copyright law.
It is harder and harder to get "what makes sense to me" to be consistent as digital technology changes the landscape, that is true. Not as a matter of law, but just your own moral sense, I'm curious if you think web-crawling for indexes for web searches (including image searches) ought to be legal. And if so, what's the difference?
As my wife keeps pointing out to me when I incessantly point out that “people on the internet are wrong about copyright”… who cares? The courts certainly don’t.
Courts have ruled against Google caching and displaying content in search timelines before. Google vs NewsCorp ultimately ended with Google paying NewsCorp license fees to index its content! But most people haven't sued because they actually want to be indexed by Google (and Googlebot respects robots.txt if they don't)
But Google search does not use images for the same purpose, and does not sell access to the images, and at present most AI models don't offer an opt-out, so the copyright implications are different (most logically, indexing and displaying an image with tacit permission and a backlink on an ad-supported web page is not the same as selling derivative works based on an image). By extension, they are not necessarily the same for StableDiffusion as they are for DALL-E, or for LLM models as they are for generative image models or licensed codebases, and copyright law has never been and probably never will be applied evenly across different jurisdictions, mediums and even individual cases.
I can't find any lawsuit between News Corp and Google. In what country? Google did agree to some kind of paid agreement with News Corp in 2001, according to news coverage at the time in part to avoid the threat of legislation in Australia that might have required Google to pay for content to index it. (No such legislation was passed in Australia, let alone the USA).
It's true it hasn't been super extensively litigated, and never made it to the supreme court. Perhaps because in fairly short order content owners realized they actually wanted to be indexed in google. And because Google (if not everyone) would respect a robots.txt to keep them out if they didn't.
The Google Books case is also probably relevant. (Which Google also, eventually, won).
There are other cases not specifically involving Google that are relevant, one for example with an index of tv news broadcasts that provided keyword search and let users view clips. The court ruled this was fair use too. Fox News v. TVEYES, Inc., 43 F. Supp. 3d 379 (S.D. N.Y. 2014).
I don't think I can find any USA cases that suggest indexing for a search engine would be a copyright violation and not fair use. Let alone against Google specifically. When you say courts have ruled against Google, can you cite any particular cases? Maybe they were from other jurisdictions than USA? Or cases relevant to this present topic even if not about search engines?
And yet, GP's description applies just as much to google indexing images. Google makes money off of a search index that copies copyrighted content without explicit permission or compensation. Still, you are right that other factors may be different; it requires an argument that goes beyond GP's, and is actually based on the law.
In the USA, where OP suit is filed, I am not a lawyer and this isn't legal advice, but I think these image AI's are going to have a pretty good case for fair use, even if it what they did counts as "making a copy" (which I think isn't totally clear). It is a very transformative use. I think convincing a court that the AI's will result in lost revenue (for the specific works used as training data, not in general for the field) would probably be necessary, but not sufficient, to overcome a fair use defense -- and I don't think that argument is even a slam dunk. And is still just one factor in fair use; the transformative nature of the use has lately been given a lot of weight in the law, and seems pretty overwhelming here.
News Corp fought Google mostly on their home territory of Australia, a battle which ended up with Google caving when News Corp succeeded in lobbying to get the law changed to be unambiguously in their in their favour (a battle AI providers should expect to face; the US isn't the only jurisdiction that's significant to them and legislation is bound to be updated for new tech). In France they got slapped with a fine of half a billion euros by an anti-trust court for not abiding by a French court order to agree fair license fees for news excerpts, and the EU's Copyright Article 13 is specifically designed to make platforms like Google News subject to EU-wide copyright law.
Google Images thumbnails were ruled infringing by one US court and overturned by another some time ago. More recently, the US also is the venue where Robin Thicke and Pharrell Williams ended up paying the Gaye estate royalties and naming Marvin Gaye a writer because Blurred Lines was apparently the result of Thicke saying he wanted to write a song that sounded like Got To Give It Up, and Williams using his knowledge of songcraft and what was distinctive about Got To Give It Up's groove to compose an original song with different words, note patterns and beat which nevertheless sounded similar, so I'm not sure there's much grounds for the absolute confidence most HNers seems to have that fair use defences will prevail in the US because of novelty and "learning".
Copyright law rulings are all over the place, and I think it's quite likely creators of such tools will be ruled infringing by one court and overruled by another, infringing in one region and OK in another, one diffusion model ruled non-infringing and another more commercially-oriented one ruled infringing, and perhaps others where either the creators or the users of the tool are deemed infringing but not necessarily both.
It doesn't matter what a lower court ruled once it's been overturned by a higher court.
As far as I know, all standing case law in the USA says that web crawling for search indexing is fair use. I'd be interested in a citation to a counter example. I've been unable to find any.
Also interested in specifics of any actual court cases involving News Corp vs Google in Australia; I can't find any references to lawsuits, only to the threat of legislation (which changes the law if passed, but I don't think it was passed in the end?)
The France thing is interesting. I know less about EU copyright law. Did Google pay the fine and... then what? Did they stop indexing news orgs in France, start paying for it, or just keep doing the infringing thing, or what? The ruling only applied to news organizations?
Has a child violated copyright when they've watched so much Disney that they can now draw a pretty good likeness of one of their characters in an original pose?
Wait. In this analogy, you think that even watching the film counts as copyright infringement? To be clear, stability have not used SD to generate an image which is infringing, they have merely trained a model which someone else could use to generate (allegedly) infringing material.
There seems to be a sockpuppet constituency desperately trying to sell "yes" as an answer to this question.
It's hard to see how the endgame of this nonsense is anything but total thought control (couched with semantic distinctions between "thought" as it occurs in your biomass and "thought" as simulated in an electronic device).
If the goal of this website was to convey information, it failed for me because I have no idea in which order to read the columns of text. Why couldn't they just make a simple page or article?
I stumbled a bit initially too, but it is in fact a reasonably readable [0] point-by-point response to the statements on the first site. They should maybe just add a landing page or a top section to explain what they're doing here.
[0] edit: on desktop, on mobile it looks pretty bad for me. shouldn't have been too hard to make this responsive I guess, just put the sections and reply below one another
It is too late. Sorry artists but it’s over. These models can be trained cheaply(compared to a model like GPT-3), run on consumer GPUs, and are open source. Even if somehow Stable Diffusion got forced to take it down, it’s already out! Will be available with BitTorrent forever. Stable Diffusion could be stopped by US copyright law, but there will be 100 Chinese and Russian startups, who will develop even better models and not give a fuck.
This is a genie that can’t be put back in the bottle. Adapt or die!
This is not mentioned enough and it's hard to think of an outcome to any legal action that would fix the things that traditional artists are asking to be fixed.
And even if there was an effective legal block on allegedly infringing models - an "IP clean" AI model would on the whole be equally disruptive. There's enough material out there in the public domain to create effective image generation models. It might not match the style of living artists very well, but that's not really the thing that's going to be the real problem.
You could say the same about any JPEG on the internet. Yet artists successfully recover licensing fees for those. It's not unthinkable that artists will be paid licensing fees for AI-regurgitated copies of their art, too.
>It's not unthinkable that artists will be paid licensing fees for AI-regurgitated copies of their art, too.
Then companies will just clean-room the thing by getting tens of thousands of developing world artists to draw things and generate training data. Combined with all the works out of copyright and those bought from existing licensors will get you probably the same quality models. That costs money but so does GPT-3 and look where we are. There is no getting the cat back in the bag, it's five blocks away with a rocket strapped to it's back.
Not only that, but all these naïve restrictions artists want to impose on machine learning would end up being awful for everyone except huge corporations.
Context: The page is a response to Matthew Butterick’s arguably frivolous lawsuits against co-pilot and now stable diffusion.
(Both websites prominently feature Butterick’s self-portrait and biography, I’m not trying to ad hominem this or anything but it appears to be his personal branding)
INAL but I think the test is does it, "harm the market for or value of X's images" so the information theory of bits and bytes doesn't really matter. I think stable diffusion, chatGPT, or CoPilot may actually fail this test - it depends on much leeway they get for the perceived public benefit and the 'transformative-ness' of this technology. What would be pretty cool would be that if the courts decided open-source = public benefit = acceptable, proprietary for profit platform = no public benefit = unacceptable without consent.
> Once again: what works? Copyright law is built around works,[18] and you have yet to show that even your clients, let alone anyone else, has works violated by infringing works that were not sufficiently transformatively different.
I messed around with Stable Diffusion and asked it to generate a lacrosse player. In the output image the crosse had a partial unmistakable unique design, too exact to be considered noise. The piece of equipment is the STX Arrow (a one-off design from a few years back). I certainly think you could argue it was not "transformatively different". Again the STX Arrow is literally the only crosse with that particular design AFAIK
This seems poorly argued. There's a lot of emphasis on technical minutiae, but the question is whether the way the model embeds and reproduces the features of the input images infringes on copyright. I would think that is more dependent on the capabilities and functions of the models, not the specific mechanism of implementation.
Also, the Jevon's paradox bit is a tangent. AI image tools may become widespread, and their use may expand the art market, benefitting artists who make use of them. That is not a relevant point, because the lawsuit is not about the legality of image-generating neural networks; It is about whether using an unlicensed image to train such a model infringes on its copyright. If it was found that it does, then AI art could still transform the art market; All that would change would be that images would need to licensed for training.
This reads like a bunch of technical arguments that have no interesting legal meaning.
You already can't avoid copyright infringement by using statistical distributions whose output is substantially similar to the original.
IE Lossy compression does not avoid infringement
This is already law.
The only question is how far does that reach and where fair use provides a defense to something that would otherwise be infringement. How transformative must something be, etc.
The basic test for infringement is access to the work and substantially similar result.
Intermediate means do not matter. You could have monkeys whose kids have genes that got passed along that enable them to redraw your work, so that they have no literal knowledge and have never seen it. If the output is the original work, its still infringement.
The article also gets wrong that court cases also have been a (reasonable) continuation of plenty of conversations, from civil rights to you name it.
I personally am not a fan of this litigation, but this author doesn't make good arguments about it
104 comments
[ 3.9 ms ] story [ 164 ms ] threadThe issue at hand is that artists had their art trained on without their permission and violating the copyright.
I've just searched the dataset to find mickey mouse, marvel and some disney characters images in the training dataset.
So then it is trained on copyrighted material, and Midjourney and especially SD are building and profit SaaS on the backs of the work without permission or compensation.
Even that alone is a valid reason litigation, and I wouldn't be surprised if Disney joins in against this or asks SD to remove them from the next model release.
Just do the following:
1. Just ask for permission to use the artists work?
2. Train on public domain images, there are plenty out there.
It's that simple.
Okay, I'll bite. Since when does copyright require permission to learn from the copyrighted material? I don't need your permission for that, and neither should an AI.
Or alternately, assuming you reject the "human-AI analogy": since when do I need permission to collect statistical information about copyrighted material? Should Disney be able to sue me for telling you how much screen time Jafar had in Disney's Aladdin, or what percentage of pixels in the first 5 minutes were a shade of blue?
Basically this is saying if you look at it, and then learn to draw you owe Disney money.
Disney would love that but it's ludicrous
(And I would not call it "learning", at least not in the same meaning of the word as for humans, but that's a more philosophical discussion.)
Fortunately, a machine isn't doing anything on its own.
It is the person who uses the machine that is doing something.
And it is the human who is allowed to use tools to create transformative work.
If I take a picture of Mickey Mouse, add 1 to each RGB value, it would still be covered under copyright.
I think the term 'learn' is doing the heavy lifting in your argument here. Specifically, your argument seems to be:
1. We use the term 'learn' as shorthand for 'fitting a machine learning algorithm to data'.
2. We also use the term 'learn' for a human looking at copyrighted material to practice their craft.
3. Humans using copyrighted material to learn doesn't make their works derivative.
4. Therefore, machine learning algorithms that are trained on copyrighted material aren't producing derivative works.
This argument holds an implicit assumption: if we use the same term for a process, it should covered by the same rules/laws. Which I think isn't very sound.
> In copyright law, a derivative work is an expressive creation that includes major copyrightable elements of an original
The key part here is "major copyrightable".
To be clear, my argument doesn't show that statistical generative approaches to art are violating copyright by using copyrighted data. It only shows that the argument "humans learn on copyrighted stuff therefore it's legal for AI to do it" isn't a sound argument.
Another argument that relies on the definition of derivative work might be able to show that you can train a model on copyrighted data without producing works that violate copyright.
"humans learn on copyrighted stuff therefore it's legal for AI to do it" - Learning can never be a violation of copyright. A work (digital or physical) can potentially violate copyright.
So there are two potential "work" here:
1. The model itself
2. Images created with the model.
I would argue that (1) can't violate copyright as it's not a derivative work
(2) may or may not violate copyright but if it does, then it does so to exactly the same degree as it would if it wasn't AI generated.
In other words "You're not allowed to plagarize copyrighted works no matter what production method you use".
We can break this into (1) and (2):
(1)
The model is a copyright violation as it contains untransformed copyrighted material. You can show this by simply asking for a picture of Mickey Mouse.
To preempt a few (obvious) arguments about what it means to be "transformed", keep in mind an encrypted zip file containing images of Mickey Mouse would not be considered transformative, even if it happens to contain images that aren't copyrighted.
(2)
> then it does so to exactly the same degree as it would if it wasn't AI generated.
This is the conclusion of an argument, but I'd need to see how you show that "it does so to exactly the same degree". Right now, it looks like you're taking that as a base assumption.
This is a strong argument and gave me pause. I think if Disney sued they could probably win.
However - examples such as this are probably quite rare. Mickey Mouse is so uniquitous that is massively overrepresented in the training and data and that has resulted in overfitting. For the vast majority of artists and almost certainly for the plaintiffs in this case, the demonstration wouldn't work. You most likely cannot produce an infringing image from a concise text prompt.
Arguably one of the outcomes of legal action might be the need for a lot of effort to remove examples similar to the Mickey Mouse example. This would be a good thing but wouldn't prevent AI image generation from broadly doing the things that the plaintiffs want to stop it doing.
> (2) but I'd need to see how you show that "it does so to exactly the same degree".
Not sure I follow? If I produce an infringing image and a court agrees it's infringing then I'm liable. It doesn't matter how the image was produced. I don't see what there is to dispute here?
For (1), I think it's handy to think of an encrypted zip file that contains a bunch of images. If you give it a valid password, it'll give you a corresponding image. If it produces an image that violates copyright, it clearly violates copyright, even if most of the images contained in it don't.
It's possible that if the model required sufficient instruction to reproduce Mickey Mouse ("create a line from (0,20) to (20,30) of thickness 2, then a line from ..."), and could not do so without that, then the argument wouldn't apply. And there's definitely a grey area once we're outside of the territory of reproducing images. But we definitely aren't anywhere close to that with these models, and looking at the "AI Safety" work, it doesn't seem possible to get these models to "behave".
> This would be a good thing but wouldn't prevent AI image generation from broadly doing the things that the plaintiffs want to stop it doing.
Yeah, I agree that the plaintiffs aren't suing solely because the model can easily regenerate their images. But I think we'll find its insanely hard to get these models to avoid producing copyrighted images. They may be forced to license the training images.
You, and only you, are liable.
If someone was distributing a password protected zip-file where if I type in "Mickey Mouse", I get a photo of Mickey Mouse, the zip file violates copyright. It doesn't matter that the underlying representation of Mickey Mouse is distorted in storage and recovered later. It doesn't matter that in order to retrieve the image, I have to go through a series of operations.
Of course, if I opened the zip file and put the image on a t-shirt, I would also be violating copyright. But the zip file itself is copyright infringing.
At this point we can just wait until the courts clarify this for you because you seem to be missing a fundamental aspect of how our legal system operates. I’d much rather go and watch football.
Without litigating if the model is "learning" in a human analogous way, instead I think the burden is on showing that the model is "substantially copying" and to an equal or greater degree than what would be needed to consider a work derivative.
And to preempt a likely counterargument, intent is not a required factor for establishing copyright violation.
If I take a JPEG of Mickey Mouse and then turn it into a PNG, it's not a "copy", as the bits are different. But it still contains copyrighted material.
You can try to argue that the bits of PNG itself aren't an image of Mickey Mouse, but rather the algorithm that reads the PNG produces an image of Mickey Mouse. But that isn't really a relevant distinction in so far as copyright is concerned.
In addition, this statement is false:
> The important distinction is that the model does not contain a copy of a copyrighted material.
It has been shown repeatedly that the model produces copies of training data. The copies are of course not stored as JPEGs/PNGs in the model, but they are retrievable from the model, given the correct password (prompt).
These models do not contain copies. One way to describe the data is they contain a statistical breakdown of the artwork, which is substantially different from a JPEG -> PNG conversion you mention.
Here's a whole paper, complete with citations:
https://arxiv.org/pdf/2212.03860.pdf
> These models do not contain copies. One way to describe the data is they contain a statistical breakdown of the artwork, which is substantially different from a JPEG -> PNG conversion you mention.
I don't understand the distinction you're making. What legally separates a "statistical breakdown" representation from a zip file representation, JPEG representation, PNG representation?
Food for thought: if I write instructions for generating an SVG of a black square, does my program contain copyrighted material (Malevich's Black Square)? You and I could argue about that, but you will probably quote more research that disproves your own point. So let's skip that.
If you're convinced that a photo of Mickey Mouse with slightly larger ears, or slightly reddish pants isn't copyright infringement, then sure, neither is any of this stuff. It would also mean that republishing copyrighted images with lossy compression algorithm (IE, JPEGs) would also not violate copyright.
I would suggest looking at the actual laws around copyright instead of relying on what you feel copyright should be.
Isn't this already well-established? For example, this image, used in The Simpsons:
https://static.simpsonswiki.com/images/d/d4/Mickey_Mouse.png
is clearly Mickey Mouse in intent, but not in a copyright infringing way.
Also, that usage of Mickey Mouse might be copyright infringing, but fall under fair use (probably parody), which is a specific defense of copyright infringement (similar to "self defense").
What I am saying is that:
1) If your model returns images which look near-identical to your training data, then any copyright infringement that applies to the training image will also apply to your image.
2) If your model can consistently return copyrighted imagery, there's little difference between explicitly sharing those images (with a password) and implicitly sharing them (via a model + prompt).
What you mean to say is that if someone just so happens to write a story that is significantly similar to another work that their intent doesn't matter... most importantly their intent is unknowable.
This doesn't mean that courts will not be interested in the intended use of specific mechanical tools such as VCRs, computers, and statistical models.
Let's tease apart this "zip file containing Mickey Mouse photos" because it seems to be from where your errors in reasoning stem.
Here is a website that gives you instructions on how to draw Mickey Mouse:
https://courses.cs.duke.edu/cps004/fall01/examples/mickey/St...
These instructions are not violation of copyright because they draw a picture of Mickey Mouse. Copyright does not protect the mere existence of an image. Copyright gives a monopoly on distributing copies of that image to the rights holder. Not the idea of the image, not the instructions that could lead to the image, not the 1s and 0s that make up the image.
As I've mentioned before, a t-shirt with the 1s and 0s that make up the zip file that contains those photos is not in violation of copyrights on those images. You'll notice that CafePress still has t-shirts with the DeCSS code for sale: https://www.cafepress.com/+,954530397
Imagine a mathematical formula that when plotted in polar coordinates draws Mickey Mouse. The mathematical formula is a fact and it cannot be copyrighted. There's another step required for copyright infringement: Someone renders the image and distributes copies of the rendered image.
https://en.wikipedia.org/wiki/Sony_Corp._of_America_v._Unive....
Take a note of the majority opinion in the Supreme Court ruling:
> The question is thus whether the Betamax is capable of commercially significant noninfringing uses ... one potential use of the Betamax plainly satisfies this standard, however it is understood: private, noncommercial time-shifting in the home.[7] [...] [W]hen one considers the nature of a televised copyrighted audiovisual work... and that time-shifting merely enables a viewer to see such a work which he had been invited to witness in its entirety free of charge, the fact... that the entire work is reproduced... does not have its ordinary effect of militating against a finding of fair use.
Stable Diffusion is also capable of commercially significant non-infringing uses. It is to be expected that the lower courts will continue to follow the precedence set by the Supreme Court that there is a clear distinction between a mechanical device capable of making copies and the distribution of those copies.
To put in the context of Stable Diffusion, this is why legal experts keep saying there are two things to consider: The model, which is almost guaranteed to be protected by fair use, and the outputs of the model, in which the users of the model are liable for any copyright infringement.
Or derivative works.
> not the instructions that could lead to the image,
And yet torrent sites have been found to violate copyright, despite not hosting any of the copyrighted content.
> not the 1s and 0s that make up the image.
Now this is plainly false. I cannot start selling a Mickey Mouse JPEG stripped straight from Disney's site.
> You'll notice that CafePress still has t-shirts with the DeCSS code for sale: https://www.cafepress.com/+,954530397
1) This is covered by a trade secrets law, and not copyright.
2) It still being for sale doesn't mean it doesn't violate copyright; only that the copyright holder hasn't litigated to stop it. Put differently, getting away with something doesn't make it legal.
Look, I think Stable Diffusion is pretty neat. But all of this "it's just math"/"it's just data"/"it also creates non-copyrighted images" stuff doesn't cut it as far as the law is concerned. If your point is that being in possession of the model isn't violating copyright, sure.
I'm talking about distributing or otherwise providing a service that uses the model to generate images. If they figure out how to get it to stop generating training data, they'll be fine. But with existing approaches, that'll be nigh impossible.
> Now this is plainly false. I cannot start selling a Mickey Mouse JPEG stripped straight from Disney's site.
You can absolutely go to the Disney website right now, download any JPEG, convert that image to a series of 1s and 0s, and then print that series of 1s and 0s on a t-shirt that you sell on your website.
Once those 1s and 0s are in a new non-digital medium they are no longer a copy of the image. Historically, the digital copy is a special case because practically there is no difference between the "expression" of a byte sequence in a digital medium and the "idea" of a byte sequence in a digital medium.
A t-shirt is not a digital medium so the rights holders don’t get to extend their monopolization of speech to a screen print of an unintelligible pattern of 1s and 0s.
Can you point me to a ruling where this form of expression is successfully defended via fair use?
Seems unlikely that a judge would agree that a simple transformation between "digital" vs "non-digital" medium avoids copyright infringement (ex: if I produce a digital image, I definitely maintain copyright over digital and non-digital forms of that image; you can't just start printing and selling posters of my image).
You could try to argue that the change of medium is transformative, but that's not a generic argument for "digital" vs. "non-digital"; that's particular to the circumstances of the case.
Finding the relevant case law is going to take me a little bit of time and it will probably require a bit of exposition so please be patient!
What are you missing about the fact that a t-shirt with Mickey Mouse on it and a t-shirt with a pattern of 1s and 0s are completely different. Is anyone buying and wearing the 1s and 0s shirt instead of buying and wearing a Mickey shirt?
You can’t say the same thing about digital copies! Is anyone buying the 1s and 0s of a JPEG in a digital medium and expecting to just have a bunch of 1s and 0s?
BTW, if anyone wants access to those 1s and 0s they only need to go to the Disney website to download the JPEG… they don’t need to buy my 1s and 0s shirt.
You’re just not thinking about this in legal terms: who has wronged who and how did they wrong them. These lawsuits are not criminal, not public wrongs, but rather private wrongs. People must show that there are concrete damages against their specific property!
What you mean is that you would prefer if Stable Diffusion was not considered fair-use by our legal system.
However, based on my more than cursory understanding of copyright law in the US, it seems more likely that the courts will find these tools fair-use than otherwise.
I think you misunderstand me. Despite my feelings about Stable Diffusion and laws around copyright, I believe the law will find in favour of the artists, assuming they have the funds/will to see it through. The law unfortunately doesn't really care how I feel about it.
Granted, they're up against some fairly large corporations that have both time and money to stall for a while, so I don't know how far their lawsuit will go.
How does that change anything at all? Take any of the plaintiffs in the Stable Diffusion lawsuit. Let’s say that their work was left out of the training data or if it was allowed to remain that the tool never reproduced training data… is Stable Diffusion any more or any less useful? Is literally anyone using Stable Diffusion with the intent to reproduce one of the plaintiff's copyright protected works?
This doesn't really have much to do with copyright.
> Is literally anyone using Stable Diffusion with the intent to reproduce one of the plaintiff's copyright protected works?
Intent isn't really a factor in terms of damages for copyright infringement. But I assume we'll figure out the extend of the damages if/when damages are awarded, since that's when they measure it out. Granted, I expect the liable parties to settle far before it gets to that; the payout seems cheap (~100k - 1m) relative to the risk (penalties for copyright infringement are per unit, and it's easy to hit millions of units on the internet).
The intentions of Sony when they manufactured the Betamax VCR were absolutely taken into consideration in the judgment.
Like, literally no one who made or uses SD knows or cares about the plaintiffs, is interested in their drawing style, and even remotely interested in any of their individual protected works. If I were the defendants I would train a model without any of the plaintiffs’ work and then show the court that the results for an astronaut riding a horse are no worse without their works.
And no, some lawyer is not going to scream out “objection, intent has nothing to do with copyright”, because, again, all that is meant by the intent is this: if Stable Diffusion output an image but your intent was not to copy another image… well too bad, you’re still infringing… this all happens where the liability for disseminating the copy and foregoing any fair use is on the person using the tool. Is using the tool worth the risk? Maybe not!
The intentions of the toolmakers are very important to the courts. A tool like a torrent website that does literally nothing to police their content and even actively promotes piracy is going to be treated differently than publicdomaintorrents.info.
> This doesn't really have much to do with copyright.
It is not specific to copyright but any private wrong heard by the courts needs a clearly defined plaintiff, defendant and property in question.
How can the plaintiffs accuse anyone of infringing on their copyrights when no one has used the tool to produce the protected works? Which of their specific works was infringed upon? Is that specific work a requirement for the tool to function?
The plaintiffs seem to be “everyone” and their property “every single digital image”, which is legally incoherent. How exactly would someone reasonably ask permission for every single digital image? Is everyone entitled to part of the settlement or just the plaintiff’s and their lawyers?
BitTorrent itself is legal. The torrent sites that were shut down were found to have basically no non-infringing utility. There were and still are other torrent sites that have non-infringing utility which are of course allowed to exist.
Yes, and so is Winzip, and the code that trains Stable Diffusion. But that doesn't have much to do with the resulting data that is used and generated by those files.
> There were and still are other torrent sites that have non-infringing utility which are of course allowed to exist.
Can you point me to a torrent site that withstood a legal challenge for copyright infringement?
The user is liable. The user is also liable for using a photocopier to create and sell copies of someone else’s work. Anyone can photocopy anything they want and plaster their own walls with that work. In all cases the manufacturer of the photocopier is not liable…
Now, if these tools consistently plagued users with being liable for copyright violations then these tools become more trouble than they are worth.
This then would put the responsibility on the creator, not the software. If I use a prompt "Mickey Mouse fighting Thanos" then the images produced are clearly derivative. But is this fair use? There's plenty of fan art out there like that. In fact, I typed that string into google and got fan artwork of Mickey as Thanos and the infinity stones are different media centers that Disney controls[0]. Making the image quite a high quality parity. I'm not a lawyer, but I don't feel like this image should be illegal. If so, doesn't that make this conversation far more complex than what you're writing here? I'd agree if I was trying to sell pictures of Mickey Mouse and claim to be Disney (and that there was no case for parody), but this does seem like a really hard line to draw and I don't think people are being honest about that.
[0] https://www.reddit.com/r/thanosdidnothingwrong/comments/9z7a...
Now, a tool that burdens an individual with a torrent of copy-written material might not be worth using, but that is separate from the data embedded in the latent space of the model itself.
I'd be careful with this terminology. These models are able to generate more diversity in data than they are able to memorize. So I just want to make sure readers do not read this as "copying and combining things to make an image" in a copy/paste manner rather than the same manner that humans do (very different mechanisms) of having a broader understanding of shapes and fuzzy concepts. But I'm also a generative modeling researcher and so I'm biased to be pedantic.
If you were to paint Mickey and sell those paintings, Disney might sue you. Why should SD (the company) be allowed to do that?
More to the point, this feels like the "have your cake and eat it" of legal arguments. If we're going to anthropomorphise AI to the extent we argue storing data in its data corpus without authorisation is just like allowing a human to look at your legally owned stuff, and reinterpreting that data in a [sometimes not particularly] lossy fashion as a set of weights and biases is just like a human learning, we should probably follow the "AI is just like a person" argument through to its logical conclusion and hold it legally responsible for not respecting copyright law if it ever produces outputs with trademarked images or uncanny resemblance to copyrighted material that aren't deemed "fair use", just like a person would be.
(tbf, it's not too hard to imagine a world in which ChatGPT says "I'm sorry, but I am not able to write stories involving Mickey Mouse, because Mickey Mouse is the intellectual property of the Disney Corporation. Respecting intellectual property rights is very important", and I'm not sure that's an improvement...)
Ah, but you are anthropomorphizing an algorithm. What is happening inside the computer you call learning, but it is just some mathematical function. Other people might not agree that this is learning, and might call it "mixing" or "copying".
Doing statistical analysis does not violate copyright, or Google would have to shut down tomorrow.
If you were instructed to counted blue pixels by somebody who used your lists for a SaaS product which could generate high fidelity representations of Death Stars, wookies and Harrison Ford (someone else probably counted the other colour pixels!) every time a customer of theirs typed in "Star Wars", quite possibly.
Models will get better and stop doing that.
Plus, regardless of poorly-enforced legal 'realities', you cannot control something once it is out in the public space. It may as well be public domain. Artists are learning first-hand that they have no control, as it should be.
I say this as an artist myself, either I release my work publicly and it will do with it what it wishes (paying me would be nice, but really difficult to reasonably expect), or I hoard my art privately and it dies when I die, unseen and unperceived.
The problem is the computer people have been dealing with bots for a while, we have tools and years of best practice to deal with it. Artists don't, this is new to them, and frankly a lot of them like to get wrapped up in emotions and do not give a damn about technical nuances -- as you correctly assert. So they will generally repeat whatever trite line sounds best. (You can see this firsthand on with the copypasted memes on facebook/reddit)
How many of these guys truly understand the issue? Or are they just the next generation of luddites trying to fight off the next printing press or camera?
That wouldn't be unusal. Collection agencies are a thing and laws can be adopted to account for AI.
Then it's pay x amount of money to use generative art.
And SAAS services can be required to collect those. Self hosted Setups can come later.
You can try -- all of entertainment certainly does with DRM and all these stupids rental models for everything. They can come up with registries and agencies to handle "collections", but all of this is toothless if I can acquire the bits and do whatever I want outside of their purview.
If I can find a way to view the artwork I can do what I wish with the knowledge and emotions that I gained from it, and there is nothing you can do to stop me. This is not so different from AI learning, the exact images are not stored in the model. Neither are they "compressed". You cannot get the same exact image out that you put in.
People need to get over this awful model of rent-seeking and collections. There are always other ways to monetize.
Yes.
> and violating the copyright.
No.
that's a personal opinion, not a legal fact. until we don't have court rulings on this matter, all legal statements are opinions. in order to have a civilized debate, we should be careful about separating opinions and facts, and clearly marking the former as such.
> The issue at hand is that artists had their art trained on without their permission
Interesting that you care about the training data but not the generated images. The only explanation I found is that you want to stop the "evil" from the root and have the models unaware of your style.
Do you believe that the style of an artist who is not included in the diffusion training set can't be replicated? No style is actually outside the scope of the model after it has learned the first few thousand styles. They become a map that will reveal every other style.
But the courts have already ruled that this is not the case. Your argument is based on what makes sense to you, but not seemingly informed by actual copyright law.
It is harder and harder to get "what makes sense to me" to be consistent as digital technology changes the landscape, that is true. Not as a matter of law, but just your own moral sense, I'm curious if you think web-crawling for indexes for web searches (including image searches) ought to be legal. And if so, what's the difference?
But Google search does not use images for the same purpose, and does not sell access to the images, and at present most AI models don't offer an opt-out, so the copyright implications are different (most logically, indexing and displaying an image with tacit permission and a backlink on an ad-supported web page is not the same as selling derivative works based on an image). By extension, they are not necessarily the same for StableDiffusion as they are for DALL-E, or for LLM models as they are for generative image models or licensed codebases, and copyright law has never been and probably never will be applied evenly across different jurisdictions, mediums and even individual cases.
One USA case from back in the day that ruled that Google Indexing (and caching) was fair use, was Field v. Google, from 2006. http://www.eff.org/IP/blake_v_google/google_nevada_order.pdf
Another USA case that ruled that Google Image indexing, and using thumbnails in search results, was fair use was Perfect 10 v. Google (2007). https://cyber.harvard.edu/people/tfisher/IP/2007%20Perfect%2...
It's true it hasn't been super extensively litigated, and never made it to the supreme court. Perhaps because in fairly short order content owners realized they actually wanted to be indexed in google. And because Google (if not everyone) would respect a robots.txt to keep them out if they didn't.
The Google Books case is also probably relevant. (Which Google also, eventually, won).
There are other cases not specifically involving Google that are relevant, one for example with an index of tv news broadcasts that provided keyword search and let users view clips. The court ruled this was fair use too. Fox News v. TVEYES, Inc., 43 F. Supp. 3d 379 (S.D. N.Y. 2014).
I don't think I can find any USA cases that suggest indexing for a search engine would be a copyright violation and not fair use. Let alone against Google specifically. When you say courts have ruled against Google, can you cite any particular cases? Maybe they were from other jurisdictions than USA? Or cases relevant to this present topic even if not about search engines?
And yet, GP's description applies just as much to google indexing images. Google makes money off of a search index that copies copyrighted content without explicit permission or compensation. Still, you are right that other factors may be different; it requires an argument that goes beyond GP's, and is actually based on the law.
In the USA, where OP suit is filed, I am not a lawyer and this isn't legal advice, but I think these image AI's are going to have a pretty good case for fair use, even if it what they did counts as "making a copy" (which I think isn't totally clear). It is a very transformative use. I think convincing a court that the AI's will result in lost revenue (for the specific works used as training data, not in general for the field) would probably be necessary, but not sufficient, to overcome a fair use defense -- and I don't think that argument is even a slam dunk. And is still just one factor in fair use; the transformative nature of the use has lately been given a lot of weight in the law, and seems pretty overwhelming here.
Of course, sometimes courts do surprising things.
Google Images thumbnails were ruled infringing by one US court and overturned by another some time ago. More recently, the US also is the venue where Robin Thicke and Pharrell Williams ended up paying the Gaye estate royalties and naming Marvin Gaye a writer because Blurred Lines was apparently the result of Thicke saying he wanted to write a song that sounded like Got To Give It Up, and Williams using his knowledge of songcraft and what was distinctive about Got To Give It Up's groove to compose an original song with different words, note patterns and beat which nevertheless sounded similar, so I'm not sure there's much grounds for the absolute confidence most HNers seems to have that fair use defences will prevail in the US because of novelty and "learning".
Copyright law rulings are all over the place, and I think it's quite likely creators of such tools will be ruled infringing by one court and overruled by another, infringing in one region and OK in another, one diffusion model ruled non-infringing and another more commercially-oriented one ruled infringing, and perhaps others where either the creators or the users of the tool are deemed infringing but not necessarily both.
As far as I know, all standing case law in the USA says that web crawling for search indexing is fair use. I'd be interested in a citation to a counter example. I've been unable to find any.
Also interested in specifics of any actual court cases involving News Corp vs Google in Australia; I can't find any references to lawsuits, only to the threat of legislation (which changes the law if passed, but I don't think it was passed in the end?)
The France thing is interesting. I know less about EU copyright law. Did Google pay the fine and... then what? Did they stop indexing news orgs in France, start paying for it, or just keep doing the infringing thing, or what? The ruling only applied to news organizations?
It's hard to see how the endgame of this nonsense is anything but total thought control (couched with semantic distinctions between "thought" as it occurs in your biomass and "thought" as simulated in an electronic device).
They'd certainly be relying on "fair use" defences, not "but it's not copyright violation if my kid learned how to do it" defences.
I stumbled a bit initially too, but it is in fact a reasonably readable [0] point-by-point response to the statements on the first site. They should maybe just add a landing page or a top section to explain what they're doing here.
[0] edit: on desktop, on mobile it looks pretty bad for me. shouldn't have been too hard to make this responsive I guess, just put the sections and reply below one another
This is a genie that can’t be put back in the bottle. Adapt or die!
And even if there was an effective legal block on allegedly infringing models - an "IP clean" AI model would on the whole be equally disruptive. There's enough material out there in the public domain to create effective image generation models. It might not match the style of living artists very well, but that's not really the thing that's going to be the real problem.
What won't happen is "x in the style of y" suddenly becoming an enforceable infringement (whether AI produced or not)
Then companies will just clean-room the thing by getting tens of thousands of developing world artists to draw things and generate training data. Combined with all the works out of copyright and those bought from existing licensors will get you probably the same quality models. That costs money but so does GPT-3 and look where we are. There is no getting the cat back in the bag, it's five blocks away with a rocket strapped to it's back.
(Both websites prominently feature Butterick’s self-portrait and biography, I’m not trying to ad hominem this or anything but it appears to be his personal branding)
I messed around with Stable Diffusion and asked it to generate a lacrosse player. In the output image the crosse had a partial unmistakable unique design, too exact to be considered noise. The piece of equipment is the STX Arrow (a one-off design from a few years back). I certainly think you could argue it was not "transformatively different". Again the STX Arrow is literally the only crosse with that particular design AFAIK
stable diffusion img: https://twitter.com/jasonmarks__/status/1570893504339116032
STX arrow: https://i.pinimg.com/originals/43/9d/e8/439de85ff04ec0f19f8f...
Are you trying to put the blame on Stable Diffusion for making you choose to publish an image with a trademark?
Also, the Jevon's paradox bit is a tangent. AI image tools may become widespread, and their use may expand the art market, benefitting artists who make use of them. That is not a relevant point, because the lawsuit is not about the legality of image-generating neural networks; It is about whether using an unlicensed image to train such a model infringes on its copyright. If it was found that it does, then AI art could still transform the art market; All that would change would be that images would need to licensed for training.
This reads like a bunch of technical arguments that have no interesting legal meaning.
You already can't avoid copyright infringement by using statistical distributions whose output is substantially similar to the original. IE Lossy compression does not avoid infringement
This is already law.
The only question is how far does that reach and where fair use provides a defense to something that would otherwise be infringement. How transformative must something be, etc.
The basic test for infringement is access to the work and substantially similar result.
Intermediate means do not matter. You could have monkeys whose kids have genes that got passed along that enable them to redraw your work, so that they have no literal knowledge and have never seen it. If the output is the original work, its still infringement.
The article also gets wrong that court cases also have been a (reasonable) continuation of plenty of conversations, from civil rights to you name it.
I personally am not a fan of this litigation, but this author doesn't make good arguments about it