AI-generated art lacks copyright protection, D.C. court says
news.bloomberglaw.com
news.bloomberglaw.com
In the review being appealed here (https://www.copyright.gov/rulings-filings/review-board/docs/...). It makes it clear that the computer scientist doing the filing was trying to argue this was a work made for hire with the author being the computer. They wanted to argue that copyright can be assigned to non humans, but that just isn’t how the law works. The summary makes it clear early that it’s just taking their word that the work had no human input and was thus purely the creation of the computer. This seems to be a a better article https://www.millernash.com/industry-news/paradise-denied-cop....
Just a nitpick that this isn't really about copyright assignment, it's about rights that arise as a result of a eligible work by a eligible author being fixed into a tangible medium. These rights aren't assigned to the author, they come into being automatically.
In this case, the ruling seems to be that there was no eligible author, and as such, no rights came into being. Or at least, the claimed author is not an eligible author: had the claim been that the human was the author, using the AI as a paintbrush, that would be a different question.
It would be another matter if a natural person authored a work and wished to assign its rights to a non-person. Which yeah, you pretty much can't do. Under most legal frameworks, ownership is for people, natural or corporate; anything that's not a people is property and property can't own property.
https://mwhlawgroup.com/can-copyright-in-software-extend-to-...
The code has no meaning without a compiler or interpreter.
The compiler and interpreter have no meaning without a CPU.
If I store x86 CPU instructions in a Postgres database, is that data or code?
It’s also a worth viewing this in light of different computing models. The main mechanism may look quite different if you’re looking at a CPU, GPU, or DSP model of computing.
I agree that both code and data are akin in terms of meaning. Transitioning from Assembly language to Python liberates us from concerns about registers and low-level intricacies. Python directs our attention to the problems we aim to solve, in contrast to low-level languages where a significant effort is spent on specifying hardware operations. High-level languages emphasize outcomes, enabling us to focus on the result rather than meticulously instructing the hardware. AI propels us further along this trajectory.
In this sense, I view a prompt and code as rather similar.
However, code is just data for a compiler.
Data has no meaning without spec. It's a quite big difference. Code has no meaning without spec too.
var a = 1 + 1 << 2;
Using a Gödel like encoding every program is also a number. So it seems that numbers that are complex enough that they are isomorphic to a copyrightable program should be copyrightable. I tend to think of the neural net weights as a different instruction set, just how Intel and ARM are also different instruction sets.
I fully well realize that lawmakers and the judiciary will be the final arbiter here (although I hope some logic will be part of the process).
Even if you consider the user input to be marginal, the issue from the link is the copyright of the program extending to the code, not that of the user.
Forget AI for a minute. Let's say I write a program to generate cool images using Penrose tiles or space filling curves or some other algorithmic process, maybe deterministic, maybe (pseudo-)random. Under that ruling, it sounds like all of the outputs of my program would also fall under the copyright of the original program, unless the creator could prove that they did substantial work in addition to what the program did.
Which then leads to the next question of: what is the "lion's share"? Does it matter if I randomly punch in some parameters and get a nice result on the first try, or if I spend hours trying different parameter combinations to find some thing that looks the way I want it to? Even if the outcome is still considered owned by the software author, is my particular parameter choice copyrightable by me, being my own creative output?
If you think about it, the modern generative "AI" systems are more or less the same thing, but where the input parameter space is something like the entire space of written language, rather than a couple drop-down menus with algorithm settings and a seed for a PRNG. Does it matter if I slap in a simple prompt and get a nice result, or if I spent hours engineering a perfect prompt?
If anything, it should be unambiguous that the prompts are copyrightable just like any other written work, being actual free-form human-readable text. So at least AI artists have that going for them.
It will be really interesting to see what happens as more court rulings unfold.
I'm being facetious of course. I suspect this technology is going to rapidly push us into radically rethinking how copyright works, but it'll be delayed some 20-30 years by incumbents desperately clinging to their hordes of existing IP.
First, there's a bunch of (non-creative) people that just want free creative assets. They don't like it, when creatives insist on being paid for the [considerable] time, effort, education and experience that goes into the creative asset.
Those people are jumping for joy.
Then, we have the creatives, themselves, who may have attended the correct schools, and have the appropriate education, to produce quality assets. They don't want people to just take their work. This looks like a side-jacking. If someone sees a nice picture that was done by a creative, then they can prompt MidJourney to produce one that looks a lot like it, for the cost of the MJ subscription. They don't particularly care (yet)[0] about copyright.
Then, we have the portfolio holders, who are often not creatives, but still hold the copyrights on a lot of creative work. Think Disney, or Universal. Their portfolios are an investment. They don't want to lose that investment.
Then, we have the crooks, who want to make cheap, crappy copies of creative assets, regardless of who owns the copyright. At the moment, there is an element of risk, as they could get caught and punished.
They are also jumping for joy.
[0] I said "yet," because one of the problems with a "wild west" copyright and trademark atmosphere, is that it is extremely easy to make "almost-like" copies, and if you don't protect your IP (or can't), then it is "up for grabs."
This is done now, but it's illegal, and can be stopped; especially if trademark is involved. I was at a Chinese restaurant, once, and was listening to the music. I noticed that there were many strains of hugely popular (and copyrighted) songs, like Scarborough Fair, and Moondance, all mashed together into a continuous stream. The owner confirmed what I suspected. He subscribed to a service that sends legal "muzak" to his restaurant. He had been mugged by ASCAP, once, and never wanted to go through that again.
I've noticed that "free everything" people often insist that everything except their own work should be "freed."
Would you consider a 3D artist generating a texture for his model to be a non-creative? What about a concept artist who mashes together AI generated images? How would you classify Corridor Digital’s Anime Rock Paper And Scissors series?
To me it seems like there’s legions of creative people who are excited to use AI in their works.
I was talking about a coder that can’t draw, using assets for icons and backgrounds, or a CEO that can’t draw or program, using AI for their entire company. I already know of examples of both.
I’m an artist[0]. When I was younger, I even considered making a living at it. I haven’t done any real artwork in decades, though.
But I’m not a particularly skilled graphic designer. I know enough to be dangerous, but most practicing designers can run circles around me.
3D artists are pretty awesome[1]. I think they will benefit from AI.
And doesn’t reward them for their contributions, thus, removing the “feeder” creativity. Soon, schools stop teaching art, because no one wants to become one, and the algorithms end up using the output of other algorithms as their input, and original creation dies.
Original creative output can be ugly and disturbing, but it is original human creation.
Tools can help this (my example is 3D modeling. When the tools were still “raw,” then only engineers could produce 3D art, and it showed. Once the tools and standards progressed, real trained artists started to use them, to produce awesome creations).
I hear, all the time, non creative people drooling over not having to pay or deal with creative folks.
This is nothing new. Just a new venue for typical human greed and laziness.
We’ll figure it out, and, I guarantee, some creative folks will figure out how to use these tools to become quite successful.
And those non creative folks are still gonna have to pay them.
This implies that without a support structure, people are not interested in creativity. Which is absurd. People are still interested in creating even if they're not compensated for it. Many people are proactively creative at great personal expense.
People painted long before it was taught in a formal setting, and people will continue to paint even if they can't use it to pay the bills (as - quite frankly - most painters do)
I know of what I speak. BTDT, got the T-shirt. One of the issues with noncreatives, is they have this illusion that creatives don't work hard, or don't "earn" their money (which can sometimes, be a great deal).
I run into that attitude on a regular. I'm routinely asked to do $20,000 worth of work for free, "because it's fun for me."
I suspect that a lot of us have dealt with that. Development of software and hardware is a creative (and valuable) effort, but people sometimes think we don't work hard at it, because our fingernails aren't cracked.
In my career, I've worked with some of the top people in the design field (and my company paid them a lot of money). They were worth every nickel.
One of the problems with being good at something, is that we make it look easy, so people think it will be easy.
I posted this a while back:
There's the old story I've heard about the retired engineer (apocryphal, I'm sure):
This engineer worked for a corporation for 40 years, and retired.
A few months into his retirement, his old company frantically calls him, begging him to come in, and help fix an issue with the system he worked on. Apparently, the new team had managed to hose it, and couldn't figure out how to fix it.
He comes in, sits down at a terminal, looks at it for five minutes, and says "Here's the issue. If you do this, it should be fixed..." He then presents an invoice for $10,000.
The beancounters go "There's no way we can pay $10,000 for five minutes work! Itemize it!"
He sits down, scribbles a bit, and presents an invoice that says:
Time to fix bug: Five minutes. $20
Six years of college, and forty years of experience, so I can fix a bug in five minutes: $9,980
There's an unfortunate culture of "Everyone [but me] should do work for free, and publish it on the Internet, so I can use it, and make money [because my own work will never be free]."
perhaps that ruling could be justification.
https://www.openculture.com/2020/03/every-melody-has-been-co...
Hell no. I can't tell if you're serious about this. You know prompt is sometimes literally a single English word right? Like literally just "pig"? If that's copyritable then it's officially the end of copyright.
Traditionally code would be written as a tool to achieve a specific output. If you automate the task of generating a particular piece of code, you still did all the work, or at least defined it pprecisely. While I don't see AI as an independent mind (yet) it seems independent enough from the person who prompts it to say the prompter doesn't own the output. I'm so 100 percent on what I just wrote, but it seems reasonable enough to say the precedent you mention may not apply.
I don't see any inconsistency in this ruling.
Actually that would be interesting hack for older people. As it could give longer period of protection than their natural lifespan.
This ephemeral fake world of contemporary society is not some sacrosanct construct, or divine mandate. There is no immutable obligation to preserve it.
This case is fundamentally not about whether art in which AI is used as a tool by a human (even if the human role is entirely in programming the AI) is copyrightable as a work of the human, it is solely about whether a piece of software can be an author under copyright law, a very different question.
https://www.federalregister.gov/documents/2023/03/16/2023-05...
This is of course up for interpretation by a judge (though I think the US copyright office refusing registration may make it de-facto impossible to get it in front of a judge, as I think it's a requirement to actually bring a suit against someone for infringement), but I suspect this is the kind of distinction that they will consider regarding this particular precedent.
The copyright office refusing registration counts as copyright registration for the purposes of register-before-suing-for-infringement.
Lel.
Its wholly a semantic game. If you say "this machine created this artwork wink wink" the court will deny you because "machines can't have copyright". If you say "I created this artwork, using aid from a computer program" they'll likely let you do just about anything.
So semantics matter. Strategic disclosure of methods will matter.
And your copyrights will be even more secure if you have an army of lawyers to say you “created this artwork”.
These are all highly subjective dials.
I think these early cases are going to get buried in cases with increasing complexity.
We don’t really know how this will fall out.
For the sake of argument, if a camera had 5 years of storage and battery life and was left running from the factory, should canon own everything it produced? Is pushing the shutter button really the most important question?
Wasn't the point that all those things were absent in that case? If the monkey indeed snatch the camera completely unexpectedly, and then proceeded to take a selfie of itself, then the photographer didn't choose the subject matter, location, angles, exposure, or anything else.
Also I don't understand why pressing a button on a camera is so different than writing a prompt and pressing a button, one clearly grants you human authorship the other is disputed.
In the case of a camera, you know how the pohotgraph will turn out. You set and positioned the camera in a specific, creative way to ensure that composition.
In the case of an AI, it's not the paintbrush, it's the painter. The prompts serve less as arms to a brush and more as instructions to a commissioned artist.
So if I stick out the camera from a corner then I don't own the copyright because I don't know how the photograph will turn out, if that's not case then I should own the copyright of an AI generated image because in both cases I have an approximate understanding of what the image would be.
This is very much like the "is electronic music real music?" debates when people can quickly throw together some presets and hit play.
TLDR it's untested in United States court, but yes, the same principles apply. You very well might not own the copyright on that street-corner photo.
There is no way of getting around the human authorship and creativity requirement. There's no loophole that AI can sneak though. If AI output does become copyrightable, it will only be because of arguments that the controlling of that AI is sufficiently creative and deliberate to pass the minimum threshold required by law and the outcome isn't a random or uncontrolled result. Saying that it's "just like pressing buttons on a camera" does you no favors, if photography was just pressing a button then it wouldn't be copyrightable.
And in fact, there was an 1884 Supreme Court challenge over photography copyright that was predicated exactly on the argument that photography was a purely mechanical reconstruction of an existing scene (https://blogs.loc.gov/copyright/2020/09/historic-court-cases...). The argument was rejected not because purely mechanical reconstructions are copyrightable, but because the court disagreed and said that photography was both an act of human creativity and that it had a controlled outcome, similar to "writing" or "engraving".
And note that a critical component of "writing" or "engraving" is intent and control. When you write down something on a page, you get the result that you intended to put on the page; you control the "words" that appear. An important question when discussing AI copyright is what the threshold is where we can say that use is actually controlling the AI. A vague prompt might not qualify, an extremely specific prompt that is itself a creative expression might? A more guided process might? (although it's not certain if this would cover the resulting image or just the prompt itself). This is all untested in current law.
Right now there's thing like ControlNet which allows you to draw a skeleton for the AI to follow. There are plugins that provide masking. You can draw stuff on top of an existing image. You can draw a sketch for the AI to follow. You can fix details in post.
So there's a multitude of possibilities available.
The interesting thing is that unless you are involved in the scene it's very hard to tell what took the user a lot of work, guidance and retouching to generate, and what popped out just like that.
Example:
Take any good digital came, switch it on, and press the button. Do you get a great photograph? No, because you haven't even removed the lens cap. Photography is not about pressing the button, it's about where you point the camera and how you configure it before you push the button.
https://www.reddit.com/r/StableDiffusion/comments/14ye2eg/co...
I could see an argument that a heavily augmented normal photo is not much different from a photo generated from a prompt. Digital zoom is virtually AI inpainting where it guesses the detail.
This case was analogous to claiming the camera as the author, and the photographer as a copyright owner only because the camera did the work as a work for hire.
Which would also fail.
Step two: Very slightly modify it. (Visual art: Apply a color filter, draw in a single element, have a different AI change some details, etc. Music: Subtly change the tempo or add one minor audio effect.)
Step three: Copyright the work as something "made by a human with AI assistance."
If challenged, step four: Claim that humans have been making works of art with all available technical means from the dawn of civilization, and that to draw an arbitrary line in the sand at assisted generation is unprincipled.
And, basically, lmao at the courts, as usual.
A classic essay:
What colour are your bits? https://ansuz.sooke.bc.ca/entry/23
> lmao at the courts
I see this a lot on HN: the idea that courts rule based on principles. It’s a strange viewpoint. There are courts of law. They rule based on law. To the extent law is messy (it is), court rulings will be messy.
There isn’t a court in the world that would make a ruling based on your, my, or anyone else’s principles.
It goes right along with the belief that capitalism is a naturalistic ideal, as opposed to a temporary state of societal organization maintained for the benefit of a few people.
Also, for the reasons you give, copyright itself is devoid of principled logic.
We can pay labourers for their time producing works. We don't need pay-per-copy.
If the AI generated portion exists nowhere else, and if it's not _obviously_ an AI-generated work, how would a court know what's generated and what's been modified?
I mean really. Do you think you’re proposing an original idea here?
Corridor Digital are getting closer to perfecting AI anime. After the first video, Rock Paper Scissors, they said they got some flak for training the AI on stills from an existing anime movie, so for the second one they hired an artist and trained it from their drawings. Well, what happens if you hire an artist that essentially draws some art work heavily inspired by an existing IP and then you train your AI model on that? Is that infringement, even though the art work you've trained the AI on is now one step removed from official source material, even if it is practically identical? What if you commission another artist to make artwork based on that artwork? And another after that? When does the signal get distorted enough to pass as not being based on pre-existing IP?
The whole thing feels like the "grain of sand" philosophy problem. What constitutes enough change to warrant something as not belonging to an IP anymore? Are we going to have to come up with some sort of image comparison algorithm that spits out a comparison percentage and set some arbitrary number as the cut off point? Or are we going to continue with time consuming and costly case by case, human judged comparisons?
The whole thing feels ridiculous.
That will absolutely always be required, because no matter how much we pretend, copyright is not a natural right, any more than capitalism is the natural economic system. You can extrapolate this to the concept of property, and by extension, ownership. The cans of Campbell's Soup don't care that Warhol copied their likeness any more than the trees we felled to build your house care if a squatter takes it over.
https://ia601401.us.archive.org/14/items/gov.uscourts.dcd.24...
This has been a problem for 75+ years. Courts just making de facto law. It’s not their role.
I type "a racoon wearing a hat" as a prompt and get back an image. It sounds like that's not copyrightable.
How about if I type that into Photoshop generative art, and then apply some filters?
Or how about I run that, and then "a walrus with a bowtie", and then "a platypus on a skateboard", and then arrange those three images together on a page?
What about if I spend literally hours prompting, and re-prompting, and combining, and layering, and cropping until I get the result I'm looking for?
These feel like important questions to me. Has anyone tested these in a court of law yet?
[0] https://www.federalregister.gov/documents/2023/03/16/2023-05...
They would be risking contravening the law. Probably not worth it.
Last week there was a book about the Maui fires in the top 100 on Amazon, which looks to have been written entirely by ChatGPT: https://www.forbes.com/sites/maryroeloffs/2023/08/17/amazon-...
As in, let's say I got AI to generate a cartoon character preparatory sketch in black penline.
If I printed that out and used it as a reference to do a "clean room implementation" of the character — do I now own the copyright, since I'm the first person to produce a copyright-eligible rendition of that likeness?
Or since it's a copy of an uncopyrightable work is the character immediately in the public domain, as if it was generated by a human & then given a really permissive license?
I LOVE being told by techbros that a human painstaking studying one thing at a time, and not memorizing verbatin but rather taking away the core concept, is exactly the same type of "learning" that a model does when it takes in millions of things at once and can spit out copyrighted code verbatim.
Because it shows how stupid they really are, but also how eager they are to scam you.
I'd argue we aren't very far from this point. Computers have already been used to make novel inventions. I think we could see an explosion in computer generated inventions if physical reasoning, reasoning about how objects move with respect to each other, continues to improve. Previously used approaches, such as genetic algorithms, have difficulty solving problems like inventing a new type of lock or clock escapement which require multiple moving parts.
If I recall there was some case in which the authorship of some picture was contested, and the judge asked the people involved to draw something right there, in court.
So the same thing would happen. They'd ask you "okay, so how did you draw it? What tools do your use? Can you show us sketches? Can you draw something here, right now?"
My "paint" is software, m'Lud. Some of it is pre-made "paint" that someone else wrote, like tubes of paint from an art-shop; some of it is my own modifications, analogous to grinding your own pigments.