An IP attorney’s reading of the Stable Diffusion class action lawsuit

https://katedowninglaw.com/2023/01/26/an-ip-attorneys-reading-of-the-stable-diffusion-class-action-lawsuit/

by spiffage • 3 years ago

91 331 3 years ago

scotty7919• 3 years ago

Stability AI has already announced that it is removing users’ ability to request images in a particular artist’s style and further, that future releases of Stable Diffusion will comply with any artist’s requests to remove their images from the training dataset.

This is incredibly disheartening. Who knows how long will it take to progress the tech to the point where anyone will be able to train and run models unrestricted without dealing with lawyer nonsense.

RandomLensman2• 3 years ago

Rent seeking by owners of AI machines is OK, but not by copyright owners?

Kiro2• 3 years ago

Yes. Abolish all copyright. Are we hackers or not?

adamsmith143• 3 years ago

I think creating art in the style of an artist is well covered by Fair Use.

healsdata• 3 years ago

I'm not sure I understand the point you're making. Its disheartening that artists can opt-out of having a computer algorithm make derivative versions of their creations?

I'm probably on the opposite side of the fence. I do find it disheartening that it's opt-out instead of opt-in. The training set should be limited to public domain and CC-0 until such a time it can comply with attribution; then other CC works could be incorporated.

t4332• 3 years ago

All art is derivative.

scotty7919• 3 years ago

It's disheartening because it's a great loss to everybody. Almost none of the people that were generating images in a style of some artist will contact this artist and pay to have an image created.

So many artists styles could have gone viral and actually bring those artists some work from the people who tried the AI commercially and got results that weren't completely satisfactory. Now barely anyone will ever have any contact with their art (relatively speaking vs scenario of virality).

Basically the only people who win are the lawyers and handful of artists that were mislead by lawyers primitive argumentation. Everybody else looses. First and foremost artists and art lovers but also AI researchers and hardware manufacturers.

smoldesu2• 3 years ago

Unfortunately, a lot of these artists opted-in the moment they uploaded their art to the internet. Once you do that, much like uploading your source code or compiled binary, it's hard to reverse the consequences. All that really happened is that the consequences changed, and a lot of people weren't prepared for it.

Yeah, it's disheartening. There's also no good way to fix it; the cost of storing copies of their art is negligible, and AI trains the same whether the material is copyright or creative commons. If you get Stability AI to omit your art, then Unstable Diffusion will be trained on your likeness. Opt-out of that one, and some guy in Nevada personally sponsors a bespoke model for making copies of your art.

So, I agree with the parent. The most tragic part is not the short-term fight, it's the long-term consequences. Artists will have to internalize what software developers realized decades ago; creating takes work, and copying is free.

horsawlarway• 3 years ago

Personally - I tend to think along the lines that we should be applying roughly the same rule of thumb here as we do for people.

People are allowed to view private art, draw inspiration and ideas from it, and execute on those to create new things.

Why should we limit AI any differently?

If the end result is too close to the original - apply the same guidelines you would for any other artist who copied your work.

Otherwise... you're not allowed copyright over a particular style (for damn good reason). While I would like to see artists retain some form of revenue, I don't think this is really the most pressing issue on that front.

matheusmoreira10• 3 years ago

Yes, it is disheartening. Technology shouldn't be held back by this copyright nonsense. Public domain? Come on. Public domain barely exists anymore with the modern multicentury copyrights.

What we need is enough computation power to run these models on our own computers, on our phones even. Then we'll be able to do whatever we want and there's nothing they can do about it.

dns_snek8• 3 years ago

I do find it disheartening that it's opt-out instead of opt-in

This is the crux of the issue for me. It's a different set of rules for AI companies than everyone else. If I started selling pirated copies of Nintendo games they would send an army of lawyers after me and this "opt-out" reasoning would not be a valid defense in court. These AI companies are trying to get away with stealing art and other content with a simple "whoopsie, we promise we won't do it again" when people demand that their own rights be respected.

marginalia_nu4• 3 years ago

It would be a spectacularly shitty world where IP protection is only granted to entities with a legal budget that eclipses the GDP of Antigua, and not to smaller independent creators.

The ends of having an useful model like stable diffusion doesn't really justify just ignoring the IP rights of tens of thousands of creators who were already having a pretty rough time making ends meet. That's just a shitty thing to do.

msla• 3 years ago

It's already the case that independent creators get their works pulled on bogus copyright claims.

aliqot• 3 years ago

Where are all these self trained artists who learned their craft in a bubble, devoid of outside influence from other artists? Is it because there's a better paintbrush now, or is it because that paintbrush is not the 'real' way?

This reminds me of the backlash against the wacom community on deviantart in the early days.

antiterra4• 3 years ago

A computer program is not a person, so the argument that stable diffusion does what a person does is of limited relevance.

haswell18• 3 years ago

Who knows how long will it take to progress the tech to the point where anyone will be able to train and run models unrestricted without dealing with lawyer nonsense.

These are orthogonal issues at this point.

The one concern I do have is that the “lawyer nonsense” (read: AI companies playing fast and loose with current laws) will stack the regulatory deck against AI technology unnecessarily - essentially because of an unforced error that brings negative attention to the technology.

Put another way, these companies are asking to have a spotlight put on them by being so flippant about copyright and ethics issues. This spotlight could have been avoided with better behavior, and the tech would still appear magical and remain one of the most impactful jumps in tech in decades.

matheusmoreira10• 3 years ago

There should be no "AI companies" in the first place. This stuff should be running on our own computers. That way they cannot set any stupid limits on it.

ok1234562• 3 years ago

It's not 'playing fast and loose'.

It's an area where there are no existing laws. We're not going to stop AI because some furry deviant art artist complains loudly online.

mk_stjames• 3 years ago

I think this is why it was very important that their first release of the model weights (the v1.4 model) was and will continue to be very important. While training an entire model from scratch requires an insane amount of data and compute right now (and will continue to be out of reach of individuals for some years I think), fine-tuning the model can be done on a consumer graphics card (they call it Dreambooth)... and I believe we will see further refinements that allow more and more useful tuning and features being built by individuals on top that original model, making it more and more powerful for specific uses. So even if future efforts change the functionality, or nerf the output in some way, there will always be people developing tools on the original weights. That cat is out of the bag.

antiterra4• 3 years ago

ChatGPT has shown that attempting to train the model to decline certain types of requests is of limited effectiveness and readily circumvented. The ability to make custom checkpoints and tools like Dreambooth will further limit these restrictions.

GaryNumanVevo3• 3 years ago

Strongly disagree, IP law (despite it's misuse by a certain mouse mascot'd company) is extremely important and protecting artists work and their livelihood.

The price floor on art commissions is already very low and AI effectively makes that cost zero, while providing zero compensation to the thousands of artists. Without their work, there's no Stability AI. From an ethical standpoint Stability is in the wrong, and from a legal one I think the class has a very strong case to recover damages.

astrange3• 3 years ago

StableDiffusion is not based on art commissions. You can search https://rom1504.github.io/clip-retrieval/ and see what kind of nonsense it usually has trained on.

quitit• 3 years ago

Interestingly it shows the tenuous nature of the plaintiffs case, even before getting into the plaintiff's large errors.

Since reasonably simplified information about SD is available and/or the plaintiff could have involved an expert to review his claims - it does raise a question if the function of the lawsuit is more about rattling chains rather than the merits of their argument. I.E. A deliberate ploy to extract a settlement.

jerf2• 3 years ago

Ultimately, this is just something that has to be solved with legislation, not a court case. It's too novel a setup for a court case to deal with under existing frameworks.

I think one issue is just that of scale. I personally tend to agree that there's something icky with just slurping up literally everyone's content, then producing a tool that will then proceed to put them out of business en masse. But proving that illegal under current law is certainly going to be a challenge.

I have not read the original complaint but it surprises me that the lawsuit doesn't have a much stronger focus on this aspect. Copyright law is very concerned about not destroying the market for a given work through infringement, but this is a case about destroying the market for entire artists at a stroke.

But that's a hard argument in court. There's no legal basis for claiming damages because the entire market itself is being destroyed.

Though I'm not sure it's any weaker than the other claims trying to be made. The basic problem is, this isn't illegal in any sense. I don't just mean "illegal in that it must be banned" but any level of gradation in between, in the licenses, in requiring compensation, in any sort of regulation whatsoever. Technology has simply outrun law again.

simiones5• 3 years ago

I think it's not going to be that hard to argue that the company is infringing the copyright of those whose images they are using. Especially once the judge is show how similar the output of SD can be to a particular artist's images with the right prompts (proving that SD has memorized a significant amount of those images).

sebzim45006• 3 years ago

Even in the strictest possible strengthening of IP law, where you need the artists written permission before feeding their data into a neural network, I think the market for artists is doomed.

Disney can train a model from every frame of their video library as well as whatever they can find which is unambiguosly public domain. Then they could hire a few hundred artists to draw whatever the model is bad at by the end of this process for finetuning.

acomjean2• 3 years ago

"The output represents the model’s understanding of what is useful, aesthetic, pleasing, etc. and that, together with data filtering and cleaning that general image generating AI companies do,2 is what the companies consider most valuable, not the training data.3"

This didn't make any sense to me. Without the curated training data (images) how are they making the models?

No matter what, putting images into your machine then selling the output generated with them and not compensating the original creators is going to be seen as problematic. Machines aren't people.

seydor• 3 years ago

They are never going to compensate the artists. It's cheaper to hire 1000 designers to make 100000 images of artistic styles they are going for

krisoft• 3 years ago

uhm. What you are proposing is compensating the artists. Those 1000 designers are the artist in question then.

It's cheaper to hire 1000 designers to make 100000 images of artistic styles they are going for

I bet that you are massively underestimating the cost of that.

gugagore• 3 years ago

I think you underestimate the scale of data these models are trained on by many orders of magnitude.

wvenable6• 3 years ago

Machines aren't people.

There's no reason why that is the significant detail. Why does it matter? If you can look at millions of images over your lifetime and faithfully reproduce famous works of art by hand, aren't you just as wrong?

EamonnMR2• 3 years ago

Humans can be trusted not to do that thing, and get in trouble if they get caught.

shagie3• 3 years ago

Machines can't create copyrighted works.

Setting aside the question of "is the model a derivative work", running the program cannot create a work that is copyrighted. Only humans (and not monkeys) can hold a copyright.

And thus, the questions are: "is generating a model based on the data set a derivate work" and the unasked question "is asking the model to generate a work in the style of {artist} a derivative work by the person asking the model?"

cma• 3 years ago

No matter what, putting images into your machine then selling the output generated with them and not compensating the original creators is going to be seen as problematic. Machines aren't people.

What about a company where you submit images and it tells you which faces are in them?

lodovic• 3 years ago

I compare it to reading a big stack of comic books and then trying to create your own, painstakingly reproducing the layout, color palettes and character styles. Or listening to (eg) Pink Floyd and then making a song that really matches their music. YouTube is full of those, it's really similar to SD images. But still not the same as sampling as that requires actually duplicating parts of the original.

The reason we have copyright for a limited time is to promote the arts, not to give people a monopoly on a specific style so they can milk it. (although that is what seems to be happening)

tshadley4• 3 years ago

"[The complaint] argues that the Stable Diffusion model is basically just a giant archive of compressed images (similar to MP3 compression, for example) and that when Stable Diffusion is given a text prompt, it “interpolates” or combines the images in its archives to provide its output. The complaint literally calls Stable Diffusion nothing more than a “collage tool” throughout the document. It suggests that the output is just a mash-up of the training data."

As noted in OP, this is an outstandingly bad definition of Deep-Neural-Networks, and the lawsuit should fail when the court hears an explanation from any competent practitioner.

However, a correct definition would make the lawsuit far more interesting, imo. Diffusion models can be compared to a superhumanly talented artist that can be cloned in unlimited fashion by anyone having the software and hardware means. How does this entity affect social well-being, how should existing laws be modified--if at all-- with the welfare of humanity in mind, etc?

matheusmoreira10• 3 years ago

the lawsuit will fail when the court hears an explanation from an expert

So how often does this happen? Somehow I'm too cynical to believe that a judge would rule against the intellectual property industry. The whole thing is based on absurd concepts to begin with, concepts that can be reduced to the ownership of unique numbers. Once a society accepts that, what difference do explanations make?

simiones5• 3 years ago

Diffusion models can be compared to a superhumanly talented artist that can be cloned in unlimited fashion by anyone having the software and hardware means.

How can you claim with a straight face that this is a better explanation of what an NN is?

An NN is simply an approximation of a multi-valued function, whose parameters are adjusted by minimizing the difference between the output of the NN and the output of the real function for a certain input. It is much much closer to "a giant archive of compressed images being used to interpolate between them" (though it's not that) than it is to a "superhumanly talented artist".

tshadley4• 3 years ago

An NN is simply an approximation of a multi-valued function, whose parameters are adjusted by minimizing the difference between the output of the NN and the output of the real function for a certain input.

Right, but that equally fits a biological NN if you zoom in that close. You'll need more than wikipedia to appreciate what deep-neural-networks are doing here, it's dimensional space that's key. What DNNs do that is similar to the human brain is that they order "concepts" in high-dimensional space. Colors, textures, shape and hierarchies of same are organized and cross-referenced with text in an incredibly complex connectome. It would be useless to memorize images with their textual descriptions as that would be horrendously inefficient/ineffective during inference. Rather, the model must do what we do and understand what makes an image a "landscape" or a "portrait" or a "cartoon". It needs to understand what is an artist's style and how to perform it on a work never before created.

"Understanding" can only mean ordering meaningless letters and pixels in multidimensional space so that they line up with human understanding (and human 'understanding', in turn, can only mean ordering meaningless sensory perceptions in the brain's multidimensional connectome such that reality turns out to be approximately predicted and controlled). The only systems that work this way efficiently are neural networks, biological and artificial.

rafale• 3 years ago

I hope the law will converge to this: As a human, I don't need a license to look and get inspired by art. But I am not allowed to feed that same data to a machine as a training dataset without proper authorization from the owner.

Karunamon3• 3 years ago

I hope the exact opposite. AI, including AGI if we ever get there, cannot be allowed to be strangled in its crib by artificially limiting the information it can learn from in the name of IP maximalism. IP law already goes way too far, the line should be drawn here.

troyvit3• 3 years ago

If you want new art you probably want some form of IP. What's the incentive for an artist if at the first whiff of success their output is overtaken and resold by technocrats with machines?

jeroenhd4• 3 years ago

The technology behind AI and AGI does not depend on copyrighted work. If the models are trained on original work, public domain works, or extremely permissively licensed work (CC0, WTFPL) then there simply is no IP conflict.

The use of including copyrighted materials in the trained model was a choice, not some obvious fact about the nature of AI. All of this could've been avoided if the data set did not include unlicensed work in the first place.

matheusmoreira10• 3 years ago

I hope this technology will become so ubiquitous that laws and "proper authorization" won't matter.

Ukv• 3 years ago

I'd rather not risk stunting progress in areas like language translation, malware scanning, DDOS prevention, spam filtering, product defect detection, scientific data analysis, autonomous vehicles, voice dictation, narration/text-to-speech engines, drug discovery, protein folding, optimization in production lines/agriculture/logistics, detecting seizures and falls, weather forecasting and early-warning systems, etc. just to let Getty Images have what they feel entitled to.

Best outcome in my opinion would be for the output to be judged on a case-by-case basis, like human works are, not for machine learning on data without "proper authorization from the owner" to inherently count as infringement.

sebstefan• 3 years ago

Stability AI has already announced that it is removing users’ ability to request images in a particular artist’s style

I hope it returns when they win and get rid of this legal bullying.

ben_w3• 3 years ago

I don't.

Information comes with many different rights: copy-right is the right to make copies; "moral rights" were mentioned in a few of my UK job contracts and that's "the right to be identified as the author of a work"; database rights are for collections of statements of fact that are not eligible for copyright but which were deemed to be worth protecting anyway for much the same reasons.

Even if copyright is totally eliminated from law by the mere existence of these AI[0], we may well retain the aforementioned "moral rights". And even if it is totally legal, there's also a strong possibility of it being considered gauche to use an AI trained on the works of those that don't like this.

[0] https://kitsunesoftware.wordpress.com/2022/10/09/an-end-to-c...

hnfong• 3 years ago

Moral rights is indeed "the right to be identified as the author of a work".

I don't think it means the author has a right to all similar styles. If I can legally ask somebody to paint me something in the style of a famous (living) artist, that person presumably having seen and studied their famous works for a while, why should I not be able to ask the AI to do the same thing?

(I understand there might be people who think even a human person emulating the style of another artist is morally wrong, but at least that's a consistent argument)

blibble• 3 years ago

it's quite common in UK contracts of employment to try and transfer the moral rights to the Employer

but this isn't enforceable, they cannot be transferred

anigbrowl5• 3 years ago

There's a big difference between 'rework my drawing to look like it was painted by Goya' and 'render this drawing in the style of Lisa Frank' or any living visual artist famed for a specific identifiable style as opposed to a particular image.

Comics are one example of an area where individual artists might develop a large body of work in a very distinctive style. You probably know what a Tintin comic (by Belgian artist Hergé) looks like. And lots of Manga artists have very specific and instantly identifiable styles. Individual artistry is a little less obvious with popular western comics because the best-known titles tend to be superhero franchises where the characters/story world are owned by a corporation and individual artists come and go.

nickthegreek3• 3 years ago

I thought many comics work with a team of artists to create their work. Largely by having 1 person create the style and other's learning to mimic it. Let alone having different people do the line work, shading, coloring, etc.

mcbuilder• 3 years ago

I don't even think it extends to this, it's simply because it's automated. I have no talent in the area, but I know that artists can copy one another's styles. You see it in talented art student master copies, and hell I'd bet most professional cartoonists could draw a page what looks pretty damn close to a series of Tintin panels when you squint.

jeroenhd4• 3 years ago

An artist's style is not copyrightable so I doubt it makes much of a difference. My guess is that showing good faith will make the lawsuit go over easier, because there's nothing illegal about paying someone to copy someone else's style (and not just a replica).

threeseed3• 3 years ago

But the artist's work itself is copyrightable.

Any use of that work without permission (and thus attribution/compensation) is the problem.

sebzim45006• 3 years ago

I think this restriction was more about trying to shut up some very vocal people on social media and less about the law.

Copying an artists style is legal in every jurisdiction in which Stability operates.

shanebellone5• 3 years ago

I've been saying this since it came out...

Stable Diffusion is equivalent to hip-hop sampling in the 80s and 90s. The outcome is obvious.

haswell18• 3 years ago

I’ve heard this argument on numerous occasions but I have never heard someone justify it or why they believe it.

Are there specific similarities that make you believe these are equivalent scenarios? Not just “it feels thematically similar”.

shanebellone5• 3 years ago

It's the closest thing to a precedent.

Hip-hop originally recorded and transformed vocals, instruments, and beats to create something new from pieces of something old. The practice occurred without permission and obviously ended up in court. Now sampling requires a licensing agreement. The additional cost has fundamentally changed the genre (over the last 40 years).

Hip-hop and tech both ignored IP rights because neither started with a legal framework and both would have found the additional cost prohibitive.

sebzim45006• 3 years ago

I think there is a pretty big difference though. In the case of sampling you can play the original against the new media and show that they are 'the same'. I have no idea how you would go about doing something similar for Stable Diffusion. "Those three pixels look different when you remove image X from the training set" is probably not a convincing argument to anyone.

shanebellone5• 3 years ago

So the litmus test for IP theft is "efficacy of obfuscation"?

Remember, input is fundamentally required. Without that dataset, Stable Diffusion delivers exactly nothing.

dns_snek8• 3 years ago

Stability AI has already announced that it is removing users’ ability to request images in a particular artist’s style and further, that future releases of Stable Diffusion will comply with any artist’s requests to remove their images from the training dataset. With that removal, the most outrage-inducing and troublesome output examples disappear from this case, leaving a much more complex and muddled set of facts for the jury to wade through.

How can this possibly be a valid good faith argument? Either they're in breach of authors' copyright which extends to every piece of art that they included in the dataset without permission, or they're in the clear and aren't obligated to respond to removal requests.

This reads like damage control to me in an effort to temporarily silence the loudest critics.

hehdhdhkf• 3 years ago

Where is the form to remove my reddit comments from chat gpt training data? Or my blog posts from gpt training data? I have a paragraph on the Internet that someone read and got an idea - I want my royalties.

These artists complaints are ridiculous, and are being made by people who don’t understand how things work.

If some other person draws a picture in their “style”, no one has to ask permission. That’s not a thing.

They either don’t understand how it works or they are just upset that a computer can make art as good as (or better than) they can in a fraction of the time.

All knowledge workers and creatives are going to face this in the future. It’s going to suck, but it would be great if we all could try to understand reality first.

dns_snek8• 3 years ago

Do you want to live in a future where artists don't make original art, musicians don't make music, book writers don't write, and so on, all because AI companies can replicate 1000 different copies in their style or merely remix it for marginally $0 cost, washed of all original copyright?

All knowledge workers and creatives are going to face this in the future. It’s going to suck

This is not a given. It's up to us and the copyright law. Real original work should be compensated appropriately unless you're proposing that we accelerate deployment of universal basic income and completely abolish copyright law.

I have a feeling you might not like the violent outcome if you effectively strip original creators of their copyright, give corporations the right to effectively generate infinite profit off the backs of their work and tell the creators (and other people whose jobs will be automated away) to pound sand when they ask how they're supposed to pay rent from now on.

EamonnMR2• 3 years ago

Where is the form to remove my reddit comments from chat gpt training data? Or my blog posts from gpt training data?

More pointedly, how do I keep my GPL'd code from spewing, license free, out of CodePilot?

yencabulator• 3 years ago

If some other person draws a picture in their “style”, no one has to ask permission. That’s not a thing.

Try making a comic book with a character that looks like Mickey Mouse and see how well that goes.

epistemer2• 3 years ago

It seems highly foolish from a marketing perspective for the artist too.

I don't see how more people copying the artist style would not increase the value of originals.

A smart artist here should promote that their style is staying in the dataset. It is as good free publicity as they will ever get.

jonhohle• 3 years ago

It depends on whether or not the “style” is associated with the original artist or if the derivatives usurp the original.

e.g. Ask 10 random people who wrote the song “Hurt” and 9 out of 10 will probably say Johnny Cash.

sct202• 3 years ago

It's not foolish at all. There's no value prop for an artist to advertise their work in the model as 1 of 400+ million. SD doesn't tell you anything about the art that inspired the output, no one will ever know the artists work was ever used, so this 'exposure' is as good as $0.

counttheforks2• 3 years ago

The models are already released. They can't retroactively censor released models.

benlivengood• 3 years ago

The LAION-5B dataset is metadata and URI pairs; all the images are publicly accessible on the Internet.

Stable Diffusion's U-Net is trained to remove noise from images in latent space, which the variational autoencoder (VAE) converts to and from pixel space. CLIP embeddings are used to improve the denoising step of the U-Net by using the correlations between human language descriptions of the pixel image to reduce latent noise. Neither the U-Net nor the VAR are trained to interpolate or reproduce images from the training set; if that happened the model would be overfitted and loss would be terrible on the validation set. The VAE is trained to produce a latent space that can accurately encode and decode any pixel image, and the U-Net is trained to remove gaussian noise from the latent space.

Stable Diffusion v2 16-bit is ~3GB of data. It was trained on hundreds of millions of images (minimum of 170M in the 512x512 step alone). That leaves a maximum of ~20 bytes per image that could conceivably be a copy, which is certainly not enough to directly reproduce either the style or contents of any individual image.

There is no artwork included in Stable Diffusion. There is a semantic representation of how images are composed of varied subjects represented in the latent space and what pixel probabilities over those subjects relate to human language phrases during decoding, and finally a method to remove noise from the semantic representation, e.g. starting with a blank or random canvas and interpreting what may be there, iteratively guided by CLIP embeddings. If you give Stable Diffusion an empty CLIP embedding you get a random human-interpretable image obeying the distribution of the learned latent space.

N1ckFG• 3 years ago

Afaik theoretically you could reproduce any image in the training set using the full weights (not a fraction of them) and the correct prompt. In practice since this is an extremely lossy process, some or most of them aren't reproducible. For this specific case, I suspect it'll come down to whether somebody in the class can pass a test like this: https://arxiv.org/abs/2212.03860

pessimizer4• 3 years ago

There is no artwork included in Stable Diffusion.

You might as well say that there's no artwork included in a .jpg, just data that can be used to recreate a piece of artwork using a carefully crafted interpreter.

wokwokwok2• 3 years ago

It’s deeply disingenuous to say the U-Net is not trained on the images because it’s trained on the latent representations.

Latents are a compressed representation of the source images that are fully recoverable.

If you train a model on a compressed jpg of an image, or on any deterministic transformation of it, you’re still training it on that image.

Any suggestion otherwise is only because someone is trying to put some spin on things.

Stable Diffusion v2 16-bit is ~3GB of data. It was trained on hundreds of millions of images…

And yet! Remarkably! It can generate pictures of the Mona Lisa!

Here’s a question for you: if you encode the process of drawing an exact copy of an image, does the pure code that implements that mean you have a copy of the image in it?

Have you encoded pixels as code?

Does that mean there’s no copy of the image?

How about a zip file full of images? It’s just a high entropy binary blob right? Yet… remarkably!!! It can be transformed into images by applying an algorithm.

I don’t know the answer, but this handwavy “it couldn’t possibly encode them it’s too small” is…

Pure. Nonsense.

Of course some part of some images is embedded in the model in some form.

Stop trivialising the issue.

The issue here is: Does an algorithm that generates content infringe copyright?

Does a black box that takes the input “a picture of xxx” and a seed and outputs a copyrighted image infringe?

You know that’s possible. Don’t dodge. Technical details about oh “it couldn’t possibly have…” are pure rubbish.

Sure it could. It could have a full resolution copy of a photo of the image in that black box.

Of all the training data? Probably not. But of some of it? In compressed latent form? Most definitely.

williamcotton23• 3 years ago

This reads like damage control to me in an effort to temporarily silence the loudest critics.

I think it is to avoid any common law wrongs related to the publicity rights of the defendants. It seems like something that a legal team would flag as an unnecessary risk for the product. Removing their names and images from the training data doesn't impact the usefulness of the model while at the same time creating a much smaller surface area for collecting subpoenas.

pessimizer4• 3 years ago

Sounds like an opt-out dark pattern. US law is unbelievably aggressive when it comes to issues of copyright, and makes copyright itself opt-out, i.e. everything you produce is copyrighted, and you have to license it in order to remove that automatic copyright. But when it comes to building these models to reproduce imitations of other people's work, suddenly copyright gets loosey-goosey.

Notice that it's a legal posture that implicitly condemns Copilot, which ignores explicitly formulated opt-outs in the form of licensing.

jacquesm17• 3 years ago

This is not just the US, this also applies to many, many other countries.

devwastaken• 3 years ago

They don't have to remove images from the training set, they're saying they're opting to do so, and using that as an argument as to why if there supposedly could be copyright infringement, they're not liable, because they allow it to be removed.

They could just as well not do anything and continue on - it's likely this case will be in defendents favor. Same as how Google can crawl the net, cache data, transform it, etc.

philipwhiuk2• 3 years ago

It's interesting the IP attorney cites The Grey Album as being an example of something that is legal, when the reality is that the case was never brought because the original artists wishes meant it was unattractive for EMI to pursue the case.

kmeisthax2• 3 years ago

The complaint includes a section attempting to explain how Stable Diffusion works. It argues that the Stable Diffusion model is basically just a giant archive of compressed images (similar to MP3 compression, for example) and that when Stable Diffusion is given a text prompt, it “interpolates” or combines the images in its archives to provide its output. The complaint literally calls Stable Diffusion nothing more than a “collage tool” throughout the document. It suggests that the output is just a mash-up of the training data.

I've seen the collage tool argument several times, and I don't agree with it. But I can understand why people believe it.

You see, there's a very large number of people who use AI art generators as a tracing tool. Like, to the point where someone who has never touched one might believe that it literally just photobashes existing images together.

The reality is that there's three ways to use art generators:

- You can tell it to generate an image with a non-copyright-infringing prompt. i.e. "a dog police officer holding a gun"

- You can ask it to replicate an existing style, by adding keywords like "in the style of <existing artist>"

- You can modify an existing image. This is in lieu of the random seed image that is normally provided to the AI.

That last one is confusing, because it makes people think that the AI itself is infringing when it's only the person using it. But I could see the courts deciding that letting someone chuck an image into the model gives you liability, especially with all of the "you have full commercial rights to everything you generate" messaging people keep slapping onto these.

Style prompting is one of those things that's also legally questionable, though for different reasons. As about 40,000 AI art generator users have shouted at me over the past year, you cannot copyright a style. But at the same time, producing "new" art that's substantially similar to copyrighted art is still illegal. So, say, "a man on a motorcycle in the style of Banksy" might be OK, but "girl holding a balloon in the style of Banksy" might not be. The latter is basically asking the AI to regurgitate an existing image, or trace over something it's already seen.

I think a better argument would be that, by training the AI to understand style prompts, Stability AI is inducing users to infringe upon other people's copyright.

anigbrowl5• 3 years ago

Great write-up. SD's removing the ability to imitate styles will probably go a long way to quell objections, though it will be interesting to see if there's a future legal split over the styles of living and dead artists. I don't imagine that anyone would object to 'autoseurat' for example.

I can see see a future dispute arising over outpainting (beginning with an existing copyrighted work) but there infringement and identity of the infringer (the user, not the toolmaker) is more clear.

Animats2• 3 years ago

That author makes the point that copyright registration (which you do online with the Library of Congress in the US)[1] is required for copyright enforcement litigation. And, quite possibly, it may be required for DMCA enforcement.

Now, that could work out. Major movie studios and recording companies do file copyright registrations and submit a deposit copy. But few others bother. It seems that you can send a DMCA takedown request without a copyright registration, but you can't enforce it in court without one.[2] This raises the question of, if you as a service receive a DMCA takedown request, should you ask the requestor to send proof of copyright registration, and if they don't, ignore the request?

[1] https://www.copyright.gov/registration/

[2] https://www.traverselegal.com/blog/is-a-registered-copyright...

rebuilder• 3 years ago

Is this requirement to register specifically a feature of the DMCA? It seems quite surprising if, as the article claims, “people who don’t have registered copyrights cannot enforce their copyrights in court.”

That would mean that the vast majority of artwork posted online is essentially free to exploit in the USA, since I’m sure most people do not routinely register their works with the copyright office before posting them.

mcbits• 3 years ago

Unless they're legally obligated to show proof of copyright registration for the takedown notice to be a valid, it would be risky to assume they didn't register it just because they didn't show proof.

Animats2• 3 years ago

Most of this revolves around the "safe harbor" provisions of the DMCA. That is, doing a takedown without authenticating the ownership of the copyright provides immunity against being sued for contributory infringement. But to actually win such a lawsuit, the purported copyright owner would have to show proof of registration.

This suggests an online process which looks like this:

* US Service provider offers web page for DMCA notices.

* Web page requests that the user enter copyright registration info.

* If user fails to provide registration info, web page offers links to various national copyright registration sites to register a copyright. A payment receipt for copyright registration is acceptable as temporary proof of registration, but must be followed up within some period of time by actual proof of registration.

* Temporary proof of registration is enough for a takedown, but the material will go back up if full proof is not submitted later.

This would put a big dent in nuisance DMCA claims. The service provider might get sued occasionally, but for big providers, it's probably worth litigating this once or twice. The companies that have valuable IP file copyright registrations. Disney will be able to show a copyright registration on all their movies.

xeyownt• 3 years ago

I don't understand how using an image as input to a model is a copyright infringement.

If the image is freely viewable (say you can browse to it), and you just look at it, are you violating any rights?

It seems that violation would only come if you would use the model to produce images that are derivative of that original image, the same way a counterfeiter would make a copy of it. Have the skill to copy is not the same as actually copying.

haswell18• 3 years ago

If the image is freely viewable (say you can browse to it), and you just look at it, are you violating any rights?

The fundamental issue with this line of argument is that it equates the process of human vision and the consequences of that with that of a computer program ingesting that image and the consequences of that.

This anthropomorphization seems like a form of deep fallacy when considering the nature and impact of AI software. In the case of "seeing" an image, the two processes could not be any more unlike each other, both in content and context.

stickfigure2• 3 years ago

Computational neural networks are modeled after biological brains. Anthropomorphizing them is not a fallacy; it's kind of the whole goal.

l33tman6• 3 years ago

The processes seem pretty alike to me (as a neuroscientist and AI researcher). Things will only move on from here, the next generation of these tools won't use a training set of 5B images and complicated month long training procedures, they will allow the "ingestion" of a style by you showing it a single instance once of a target image and it will immediately know the style (just like a human artist would).

I'm not putting any weight here on what is good or bad for society, but relying on that humans somehow work in a completely different way from where AI is and is going is not going to help.

I do think it will take longer for the AIs to know all about human contexts though, so the pairing of human AD + bulk-gen AI seems to me to be an obvious near-term tag team that's hard to beat.

LeifCarrotson• 3 years ago

This anthropomorphization happens all the time in the other direction: Software actors are seen reading license plates, wiretapping connections, checking speed limits and red light compliance, monitoring uploads for copyright infringement, issuing takedowns, and otherwise acting as legal entities. Government actors are constantly allowed to do things because those rights would be afforded to an individual policeman or other human agent.

I agree that this is a dangerous fallacy. Something that legislatures and culture have agreed is fine for a human to do - limited by human scaling, memory, and skill - may not be fine for a computer to do.

counttheforks2• 3 years ago

If the image is freely viewable (say you can browse to it)

This doesn't mean anything. If an unsecured SSH server is connected to the internet that lets anyone who connects to it in and gives them a root shell, it is still illegal to 'hack' that machine. The law cares about intent, not technicalities.

edit: Since HN decided to break with "You're posting too fast. Please slow down. Thanks." again, banning me from replying: This is obviously just an example intended to show that the law cares more about intent than technical measures.

@dang Calm down dude.

verdverm• 3 years ago

Copyright and intrusion are different areas of the legal code and have different interpretation and allowances. For example, copyright typically has a fair use exclusion for infringement

sandworm1012• 3 years ago

> If the image is freely viewable (say you can browse to it), and you just look at it, are you violating any rights?

If I read Harry Potter, then turn around a write a book about a wizard with a z-shaped scar? Who works at a school for wizards? With a pet owl? Who is an orphan? At some point I have started to violate intellectual property rules. (Ignoring all the Harry Potter material that was itself lifted from prior public domain art.)

AI systems aren't just reading, they are generating material based on the stuff they have read. They and the people controlling them have to abide the copyright rules just like any other "author".

l33tman6• 3 years ago

That wasn't the question. The question was if the learning process itself is violating any existing copyright laws.

mjhay2• 3 years ago

Human artists/writers are influenced by each other all the time. I really don't see how it is fundamentally different. Most of Harry Potter is derivative of previous fantasy work itself. Nothing is made in a vacuum.

https://tvtropes.org/pmwiki/pmwiki.php/Main/WizardingSchool

km3r6• 3 years ago

Humans aren't just reading, we're constantly updating our brains neural nets. Both the AI system and brains may be capable of writing a copyright infringing rip off of Harry Potter, but the ability to do so isn't infringement only actually doing so.

williamcotton23• 3 years ago

If the image is freely viewable (say you can browse to it), and you just look at it, are you violating any rights?

This isn't the kind of question that the lawyers of the defendants are going to ask the court.

They'll more likely ask if it isn't clearly fair use similar to Sony v Universal and Authors Guild v Google and then present evidence of significant non-infringing commercial use.

It seems that violation would only come if you would use the model to produce images that are derivative of that original image, the same way a counterfeiter would make a copy of it. Have the skill to copy is not the same as actually copying.

Yeah, that's basically how the courts see it these days although for a different reason. They don't ask questions about skills or work or anything like that. They ask questions like, "is this supposed infringing work a replacement in the market for the plaintiff's work?".

The deeper questions about what the hell anyone meant by the words in the Constitution about Copyright wait for the highest courts to get involved, which is where we got this nice division between tools and what the tools are used for which allows for innovative fair use of copying other people's protected works with tools like VCRs, online book search and large language models.

jacquesm17• 3 years ago

They'll more likely ask if it isn't clearly fair use similar to Sony v Universal and Authors Guild v Google and then present evidence of significant non-infringing commercial use.

Those were not cases about 'generators' but about 'aggregators', a completely different class of application.

BeefWellington• 3 years ago

Authors Guild v Google and then present evidence of significant non-infringing commercial use.

The problem as I understand it is in all the likely "precedent" cases for this, what was being done with the scraped data was in some identifiable way different than the purpose of the source data itself. Authors Guild v Google for instance, the argument was that Google wasn't reproducing whole texts, it just used that data to make the texts searchable. Meaning the purpose of the consumption of the text by Google was to essentially make a searchable index, rather than to reproduce a book, and thus that isn't harming the authors.

In this case, it would seem a very key difference is that this is Art being consumed and Art being produced, with no different purpose.

pessimizer4• 3 years ago

I don't understand how using an image as input to a model is a copyright infringement.

Look at the extreme case, then. What if that one image is your only input, and your output is identical to it? What if your output is your input reflected over the x-axis? What if your output just crops the input? What if your output is your input cut into irregular pieces and randomly rearranged? Which outputs violate copyright?

Slightly less extreme: suppose your input is two images, and your output is those two images next to each other in a single image? Or your output is the second image, reduced in size and placed in the center of the first? What if both of the inputs are human figures, and your output is to cut out the face and hands of one image and put it onto the other?

images that are derivative of that original image, the same way a counterfeiter would make a copy of it.

Only one of these outputs are anything a counterfeiter would do. Are any of the others copyright-violating?

threeseed3• 3 years ago

If the input argument were true then what about apps like Adobe Lightroom.

Would they be able to use your photos for Adobe Stock without permission ?

majormajor• 3 years ago

It seems that violation would only come if you would use the model to produce images that are derivative of that original image, the same way a counterfeiter would make a copy of it. Have the skill to copy is not the same as actually copying.

With this sort of model's "creation" process, is something close to everything it generates derivative of everything it ingested, since had you ingested a different set of images you'd presumably have a different model with different weights?

That's kinda sorta analogous to human creation, but a human can much more actively choose what to think about, what to ignore, what to filter out.

The human process involves an explicit creative judgement step that I don't think the image-generation-by-model process can - and that creative transformation is key, legally, to a derivative work being able to itself be copyrightable and to not be infringing.

lofaszvanitt2• 3 years ago

Think of it as this way:

in order to create 5 very different illustrations you need to talk with 5 people. in the end 5 people will get money when they finish with their work.

an AI consumes these artists past output and instead of paying to these artists it will gather income to the owner. So by using the output of 5 people who have spent decades on perfecting their craft, the AI generates income by stealing their work, and the money flows to the owner only, who doesn't give back anything to these people.

so in essence AI in this form kills income stream for humans, since it gives back nothing.

theRealMe3• 3 years ago

Throughout history almost all skills have been learned/copied from other people. Especially things like art are learned by studying previous work.

What specifically is the defining reason that people can learn by copying other peoples styles but ai cannot?

Are we supposed to halt technological progress to avoid antiquated job destruction?

consumer451• 3 years ago

A lawyer who works on YouTube channel Corridor Crew posted a decent breakdown on this lawsuit recently as well:

https://news.ycombinator.com/item?id=34479857

mensetmanusman• 3 years ago

“LLMs are illegal because anything they see is owned by other people”

The Disney protection act rears its head…

layer8• 3 years ago

future releases of Stable Diffusion will comply with any artist’s requests to remove their images from the training dataset.

How does this work? Do they retrain the model from scratch every week? Or is it somehow possible to retroactively remove specific training-set items from the already-trained model?

cycomanic2• 3 years ago

I don't really understand the argument about danger mouses grey album being different from just a random "mash up" because the artistic merit behind the grey album. Sure the grey album is likely much more pleasant to listen to and would likely be considered worthy of copyright itself, where a random mash might not be. That doesn't change the fact that danger mouse had to ask permission to use Jay Zs and the Beatles work (and likely had to pay), or otherwise would have violated copyright. So how is that argument relevant. Nobody is arguing that composing images via stable diffusion prompts (like making some collage) is not a creative process. The argument is does one have to have permission/licence of the original creators.

gcoakes• 3 years ago

Am I the only one who thinks this just isn't defined well enough to be decided by the judiciary? It should be legislated. My opinion is that ML training should be distinctly different from human learning.

jen202• 3 years ago

I agree that it should both be legislated, and substantially different to human learning. In the US (or UK) however, most legislators are such intellectual lightweights that they have no hope of grasping even the basics of what they are legislating.

An IP attorney’s reading of the Stable Diffusion class action lawsuit

Related Stories