Autonomously AI generated art cannot be copyrighted.
this post was submitted on 20 Aug 2023
493 points (96.8% liked)
Technology
59111 readers
3339 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
Just thinking out loud: how would this impact AI-generated videos, or stuff like AI-generated actors and AI-written scripts? Does this suggest that stuff made by AI would, by default, belong to the public domain? If true, that could do quite a bit in forcing the movie studios to get off their asses and bring them back to the negotiating table with the actors and writers.
The ruling says the AI itself can't hold a copyright, but humans using them can still be copyright holders of any qualifying works.
So we'll hear this case been wrongly cited forever. AI art is often human guided and sometimes involve editing and adjusting. Rarely ever I get something good the first prompt. But I bet people will say lol no copyright.
Eh I think it would have been worse if this guy won. To my knowledge he was trying to get the AI to be considered the author and then himself to be the owner of the copyright via the "work for hire" clause. As I understand it that would have been catastrophic. It would have likely meant that anything users prompt from these generators would automatically be the copyright of the people running the AI.
The process you describe could likely still be protected under this ruling since there's human involvement in the selection of output to use and the altering of it afterward to fit whatever creative vision the person had. If this had won a person doing that it seems would at best be making a derivative work and still not be able to protect it.
If they had won then copyright trolls like prenda law would have flooded the internet with AI crap and sued everybody that ever make anything resembling any of their outputs, claiming copyright infringement.
This ruling strongly throttles the ability of copyright trolls to use ML that way because the defendant can much more easily argue they are producing far too many works (and with to many ML-ish obvious errors) to be human made, thus no copyright protection and then you don't even have to prove you didn't copy it anymore (but of course you should still try to argue both when you're the defendant).
Sidenote - derivative works can be protected separately if the addition itself holds creative height. Your copyright only covers your own addition.
Yeah you're right about the derivative works thing. I glossed over it mostly because that seemed pretty much useless to me in the realm of visual art. But I suppose there are a few scenarios where that ability to get protection is meaningful.
Either way, I can't see a way that this would have been good for anyone if this guy got what he wanted.
Please correct me if I am misunderstanding something here, then: Doesn’t the ruling here state that the human using the AI tried to apply for a copyright listing himself as the copyright holder and the AI as the author that worked on a commission for him, which is what was denied? Or are you saying that the reason it was denied is because he listed the AI as the author?
Here's the relevant part from a different article:
Thaler, who listed himself as the owner of the copyright under the work-for-hire doctrine, sued in a lawsuit contesting the denial and the office’s human authorship requirement. He argued that AI should be acknowledged “as an author where it otherwise meets authorship criteria,” with any ownership vesting in the machine’s owner. His complaint argued that the office’s refusal was “arbitrary, capricious, an abuse of discretion and not in accordance with the law” in violation of the Administrative Procedure Act, which provides for judicial review of agency actions. The question presented in the suit was whether a work generated solely by a computer falls under the protection of copyright law.
Got it, thanks!
Without human authorship you cannot have a copyright. If something has no copyright protection, it is public domain.
However, the public domain is not viral. A work made with public domain elements can itself be copyrighted. However the copyright will only protect the creative expression added to the public domain work. Everyone else is free to make their own works from the public domain elements.
But...
Holders of public domain works are NOT obligated to publish or make available public domain works in their possession.
So if you use AI generation as part of your process you still have a valid copyright. Unless the audience can extract the unprotectable elements from your final product, you have the same copyright protection as a fully human produced work.
This ruling only applies to fully AI produced works. Using AI to modify a human performance to look or sound like someone else, still copyrightable. Human filming from an AI script, still copyrightable except for the script itself.
If AI makes the final output is where there's trouble. AI's aren't human, their expression isn't copyrightable. The prompt you give the AI is most likely factual rather than creative, which would make that uncopyrightable as well.
Copyright protects humans' creative expression, and nothing else.
This was a great summary, hope more people read this.
If AI makes the final output is where there’s trouble. AI’s aren’t human, their expression isn’t copyrightable. The prompt you give the AI is most likely factual rather than creative, which would make that uncopyrightable as well.
I am not convinced by that. I don't think any lines have been drawn on how much creativity you put into a prompt for the resultant image to be considered copyrightable. This case is about someone trying to get the AI to claim copyright, and have that transferred to him as the owner of the AI. Which is like the cases where someone tried and failed to claim copyright over an image of a monkey that the monkey took because it was taken on his camera. He had no creative input into the shot so the image was not copyrightable.
You could condisider the AI like a camera - you have control over its input, what you point it at, the lighting levels, even creating the scene you point it at. All of which are creative elements. You did not actually create the image - light hitting a film did that. You just set it up to capture the picture you wanted.
So I could see the prompt being similar to setting up a shot for a camera and the AI being like a camera. With enough creative work being put into setting up a good prompt I can see a valid claim for copyright being made - though I am not aware of anyone testing this out i court yet.
load more comments
(6 replies)
I pray it means that Disney is screwed on their plans to make their productions fully autonomous and fire their writers, animators, other artists, line producers, ... I could go on, but you get the idea.
I don't need confirmation that they are working towards that goal, it is the inevitability of their existence and greed. All profit, no overhead.
Good questions. Might be the case. I really appreciate that this decision strengthened the authorship as originally a human authorship. Otherwise many other ownerships might be taken over by computers. With ownership reliability and consequences came hand in hand, but a computer gives a shit about consequences.
IMO more importantly that copyright of the work does not go to the owner of the AI systems that created it. Which strengthens the copyright claims for anyone that came up with a creative enough prompt that was used to generate the image. And drastically weakens it for those running the servers or training the AI.
My god....is this an island of reasonable thought in a sea of insanity? Good precedent to set, otherwise every moron would copyright every image they made
That's what we were doing before Ai lol
I hope all AI generated everything can't be protected.
Nah, fuck that. Writing aides and stuff work wonders to break me out of bad habits and writer's block. I'm 200 pages into a thing because the LLM I use is like a buddy who takes your writing and goes, "yeah, and--!" And keeps you going. It's my content in my style based on my writing, idiosyncrasies, and colloquialisms. It's augmented by an LLM.
This is how it should be. A tool to help people be more productive and effective at their jobs.
So, no change then. This is as it has always been. You can take AI elements chop them up, recombine them, and have copyright over the result, but you can't say "Show me a picture of waffles" in your prompt and expect the resulting waffles to be copyrightable.
That is not even what this case was about - it was about some guy that created an AI that used that AI to generate an image with no real inputs - then claiming that the AI was the author and the copyright should transfer to him as the owner of the AI.
No precedence was set on how much human creative input was needed for an AI created image to be considered copyrightable. Only that you need to be a human to copyright something. Which was already set by previous cases where animals took a picture of things/themselves. No you don't get copyright if you own the camera, just like you don't get copyright for owning the AI.
I am not aware of any cases yet which start to set the bar for what is considered enough human creative work for the content to be considered copyrightable.
load more comments
(1 replies)
This title is massively misleading. There is no ruling that says AI-Generated Art cannot be copyrighted at all. This case is about someone who filed for copyright listing the AI as the author and that the copyright should be transferred to him as the owner of the AI.
Plaintiff Stephen Thaler owns a computer system he calls the “Creativity Machine,” which he claims generated a piece of visual art of its own accord. He sought to register the work for a copyright, listing the computer system as the author and explaining that the copyright should transfer to him as the owner of the machine.
The claim was rejected on account of you can only claim copyright if you are human and an AI does not count as a human - so no AI can claim copyright over a works.
The Copyright Office denied the application on the grounds that the work lacked human authorship,
And he explicitly stated he gave no real input into the work.
Plaintiff requested reconsideration of his application, confirming that the work “was autonomously generated by an AI” and “lack[ed] traditional human authorship,”
But that does not mean AI generated work is uncopyrightable like the title claims - only that enough human input needs to be present to be able to claim copyright over any works. We have yet to decide on how much input is required for someone to claim copyright over an AI generated image, which the case clearly states:
Undoubtedly, we are approaching new frontiers in copyright as artists put AI in their toolbox to be used in the generation of new visual and other artistic works. The increased attenuation of human creativity from the actual generation of the final work will prompt challenging questions regarding how much human input is necessary to qualify the user of an AI system as an “author” of a generated work
So AI generated work is not uncopyrightable by the own conclusion of this case. Making the title of this article a complete lie. More cases will likely be done to draw the line as to what really counts as enough human input - this case was not one that does that. Only confirms that non-humans cannot claim copyright over an image. And that you need enough human input for a work to be copyrightable.
This is the best summary I could come up with:
United States District Court Judge Beryl A. Howell ruled on Friday that AI-generated artwork can’t be copyrighted, as noted by The Hollywood Reporter.
To contrast, Judge Howell noted a case in which a woman compiled a book from notebooks she’d filled with “words she believed were dictated to her” by a supernatural “voice” was worthy of copyright.
Judge Howell did, however, acknowledge that humanity is “approaching new frontiers in copyright,” where artists will use AI as a tool to create new work.
She wrote that this would create “challenging questions regarding how much human input is necessary” to copyright AI-created art, noting that AI models are often trained on pre-existing work.
His attorney, Ryan Abbot of Brown Neri Smith & Khan LLP, said, “We respectfully disagree with the court’s interpretation of the Copyright Act,” according to Bloomberg Law, which also reported a US Copyright Office statement saying it believed the court’s decision was the right one.
Sarah Silverman and two other authors filed suit against OpenAI and Meta earlier this year over their models’ data scraping practices, for instance, while another lawsuit by programmer and lawyer Matthew Butterick alleges that data scraping by Microsoft, GitHub, and OpenAI amounted to software piracy.
The original article contains 386 words, the summary contains 201 words. Saved 48%. I'm a bot and I'm open source!
In this case, couldn't an artist simply not disclose that they used AI for things like script writing or character creation? It would be on the public to figure it out, wouldn't it? It's not necessary to prove that you didn't use AI in creating the works, is it?
Kinda, the thing is they are transferring the standard used for stuff like photography where they require a certain creative height which is required to be supplied by a human to apply copyright.
They used the same example of cameras which I've used myself - it's not the act of pressing the button to produce a picture which gives you copyright over the result, it's instead the act of expression by choosing a motive which gives you copyright.
Similarly with ML tools, if you used a mere list of facts as inputs (lists of facts are not copyright protected, so eg. recipes are therefore not protected) then your output doesn't contain human expressed creative height, you effectively just rolled a dice a few times when running the ML model.
Now if nobody notices then you can probably claim copyright and get away with it, similarly to accidentally taking a great photo when you didn't intend to press the button. But somebody else might try running a few prompts in the same ML model and argue your work looks like it was derived from it with a much too simple prompt to get copyright protection.
If you want copyright protection for your own ML processed works, make sure your prompts are actually creative, similarly to what it takes for a poem to be considered enough for copyright protection.
I didn't realize you could copyright art like that.
Any time you make something, you have a copyright on it.
You have a copyright on the comment you just made, for example. And I have a copyright on this reply. It just magically happens once you create the work.
You can give your copyright away (for example, allowing Lemmy to publish your work on other instances or show it to others). You can also sell your copyright; when a publisher buys a book from an author, they actually buy the copyright to the words the author wrote (and thus the author loses their copyright over the work).
This goes beyond just words - pictures and whatnot have the same inherent copyright.
Any time you make something, you have a copyright on it.
I don't think that is quite true. I think there is a minium bar of human creativity needed for copyright to apply to something. If you accidental knock some paint over onto a bit of paper you do not get copyright over the result. But if you pick some paint, and intentionally throw it at a canvas in deliberate motions you have a much stronger claim of copyright over that work.
The work i believe also needs to be big enough to be able to claim copyright. A single sentence might not be enough, like how you cannot copyright a single cord in music.
How does this work for using ai generated art as part of larger projects e.g. games development? Is the game still copy rightable? Are parts of it protected but others not?
The ai generated portion does not have copy right protection. This also applies within an image. So for instance in an ai generated building image with a human created character in front. People couldn't copy the character but could use the background.
How will the copyright office know what's ai generated and what's not
They'll have to find the tools that will help them detect AI works. However, the current standard they've set is that once they learn its AI generated the work is no longer protected under copyright law.
AI generated portions of things would have copyright per the article. Only wholly-AI-created content is non-copyrightable per this ruling.
The title is a lie if you actually read the case. AI work can be copyrighted if there is enough human input into the work. This case was just about trying to get the AI to claim copyright (which it cannot do as it is not human) and transferring that right to the owner of the AI.
As for cases where the AI work is not under copyright then I believe it would be consider public domain? And thus we already have rules for how that works can be used as part of a greater works.
Also, is there even any need to tell people that parts aren't copyrighted? That'd be pretty tedious to do ("texture on model pot_interior_clay_2_cracked is not copyrighted"). But if a game has a mix of copyrighted and non copyrighted media, that basically means nobody can use the non copyrighted parts because they simply can't identify which parts those are.
I suspect there's no need to tell people. After all, mixed media is already a thing. I can make a copyrighted video, for example, in which I quote some Shakespeare. The Shakespeare quote isn't copyrighted, but the rest is. I've never seen any kind of copyright notice mention this.
So the net result might not be any different. Just if you steal assets from something, they might have a harder time identifying if they have a case of copyright infringement. Only if something was entirely AI generated would things likely change much. Though there's also some weird edge cases. Like what if a human makes a 3D model but an AI textures it. 3D models are basically never used without their texture. So what's the copyright implications of using videos of this textured model? Perhaps something for a very expensive legal case to figure out?
So what’s the copyright implications of using videos of this textured model?
If the textures are not copyrightable then they are in the public domain are they not? If so the same rules apply to any public domain works and I believe you can use them however you want to in your own works.
Technically no. But law is all about interpretation and it can be easy enough to hide depending on what stage of the project it’s used.
TBH, this really only affects people who thought they could gut their entire writing team or artists and release content solely done by AI.
It’s unlikely to affect places that use it for things like mood boarding or other menial tasks.
This has potentially interesting implications for AI generated source code and "copilots"
Howell wrote that copyright has never been granted to work that was “absent any guiding human hand,”
Plaintiff develops and owns computer programs he describes as having “artificial intelligence” (“AI”) capable of generating original pieces of visual art, akin to the output of a human artist.
If he developed the program, that sure sounds like a "guiding human hand" to me. I think his real mistake was trying to claim it as a work for hire with the AI as the author, rather than it just being a tool.
Copyright protects creative expression. Inventions are covered by patents. The guiding human hand would need to file with the patent office rather than the copyright office to protect their art algorithm.
When the copyright office sees the output of the art algorithm, they see an image that is not copyrightable due to a lack of human expression, a prompt that is uncopyrightable due to it being a factual list of things the image should contain, and an AI that is not even governed by their laws.
The ruling is unsurprising.
Yes, the protection of the software and the images it creates are separate, but that's missing the point. What protections the software does or doesn't have are irrelevant to the question of whether or not the images are covered. By developing the software, he determines how it functions, which influences the final product that it outputs. That would still be the case even if the software weren't covered by IP of any kind at all.
I'm pretty sure that he wanted to go this route so he could have automatic copyright ownership of literally anything people using his AI generator prompted from it. There's already ways that artists can take AI output and pretty easily make it something that can get copyright protection. It really seems like he was just angling to own by default anything that is generated using his AI.
load more comments
(1 replies)
Well this is going to the Supreme Court I guess.
view more: next ›