AI-Created Art Isn’t Copyrightable, Judge Says In Ruling That Could Give Hollywood Studios Pause
More than 100 days into the writers strike, fears have kept mounting over the possibility of studios deploying generative artificial intelligence to completely pen scripts. But intellectual property law has long said that copyrights are only granted to works created by humans, and that doesn’t look like it’s changing anytime soon.
A federal judge on Friday upheld a finding from the U.S. Copyright Office that a piece of art created by AI is not open to protection. The ruling was delivered in an order turning down Stephen Thaler’s bid challenging the government’s position refusing to register works made by AI. Copyright law has “never stretched so far” to “protect works generated by new forms of technology operating absent any guiding human hand,” U.S. District Judge Beryl Howell found.
The opinion stressed, “Human authorship is a bedrock requirement.”
The push for protection of works created by AI has been spearheaded by Thaler, chief executive of neural network firm Imagination Engines. In 2018, he listed an AI system, the Creativity Machine, as the sole creator of an artwork called A Recent Entrance to Paradise, which was described as “autonomously created by a computer algorithm running on a machine.” The Copyright Office denied the application on the grounds that “the nexus between the human mind and creative expression” is a crucial element of protection.
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.
“In the absence of any human involvement in the creation of the work, the clear and straightforward answer is the one given by the Register: No,” Howell wrote.
U.S. copyright law, she underscored, “protects only works of human creation” and is “designed to adapt with the times.” There’s been a consistent understanding that human creativity is “at the core of copyrightability, even as that human creativity is channeled through new tools or into new media,” the ruling stated.
While cameras generated a mechanical reproduction of a scene, she explained that it does so only after a human develops a “mental conception” of the photo, which is a product of decisions like where the subject stands, arrangements and lighting, among other choices.
“Human involvement in, and ultimate creative control over, the work at issue was key to the conclusion that the new type of work fell within the bounds of copyright,” Howell wrote.
Various courts have reached the same conclusion. In one of the leading cases on copyright authorship, Burrow-Giles Lithographic Company v. Sarony, the Supreme Court held that there was “no doubt” that protection can be extended to photographs as long as “they are representative of original intellectual conceptions of the author.” The justices exclusively referred to such authors as human, describing them as a class of “persons” and a copyright as the “right of a man to the production of his own genius or intellect.”
In another case, the a federal appeals court said that a photo captured by a monkey can’t be granted a copyright since animals don’t qualify for protection, though the suit was decided on other grounds. Howell cited the ruling in her decision. “Plaintiff can point to no case in which a court has recognized copyright in a work originating with a non-human,” the order, which granted summary judgment in favor of the copyright office, stated.
The judge also explored the purpose of copyright law, which she said is to encourage “human individuals to engage in” creation. Copyrights and patents, she said, were conceived as “forms of property that the government was established to protect, and it was understood that recognizing exclusive rights in that property would further the public good by incentivizing individuals to create and invent.” The ruling continued, “The act of human creation—and how to best encourage human individuals to engage in that creation, and thereby promote science and the useful arts—was thus central to American copyright from its very inception.” Copyright law wasn’t designed to reach non-human actors, Howell said.
The order was delivered as courts weigh the legality of AI companies training their systems on copyrighted works. The suits, filed by artists and artists in California federal court, allege copyright infringement and could result in the firms having to destroy their large language models.
In March, the copyright office affirmed that most works generated by AI aren’t copyrightable but clarified that AI-assisted materials qualify for protection in certain instances. An application for a work created with the help of AI can support a copyright claim if a human “selected or arranged” it in a “sufficiently creative way that the resulting work constitutes an original work of authorship,” it said.
Don't post the entire article in the OP, please. You'll end up getting C&D's sent to your instance admins if publishers keep seeing this, because it's - ironically enough in this context - copyright infringement.
Just post a snippet to stay within fair use. Don't ruin Lemmy for all of us over something so silly.
Ok, my bad
What if one feeds the entire article into an LLM and has that rephrase it? Is it derivative then?
Well you can’t copyright what the AI wrote
Only if you say it was written by an AI, that's the lesson here.
Other journalists websites do this all the time now, and claim authorship. If they can get away with it, I don't see why we can't.
Only if it introduces biases and errors like a normal person would.
That's honestly not a bad idea. I might start doing that next time I post a link, myself!
That's a normal newspaper's article. Most articles (non-opinion articles) are rephrasing of press releases from press agencies
Only if you add flying turtles and snarky goblins into the article.
Or, even better, use AI to generate a tldr.
Isn’t that what the autotldr bot basically does?
Exactly
@Chozo @ugjka how’s that if the article is actually linked on the post.
Linking to it is fine, but OP had copy/pasted the full text of the article into the body of the post. It looks like he's since edited it out.
It doesn't happen too often, but I've seen some websites get in trouble for doing that.
That's great! It means artists can continue to use AI art for projects they don't intend to sell, and Hollywood, which already has too much power, still relies on others.
Artists can still make money and copyright their stuff. You just can't use exclusively AI to create the images. Cleaning up an AI generated image count as artistic work. Color correct, add missing fingers, make the eyes point the same way, remove background monstrosities. It all adds up.
Unfortunately this also goes for Hollywood. They can generate the bulk of the work and have one guy do the editing and suddenly they own the edit.
The real losers in this are the people that generate images with no modifications and post it as is while pretending that they are doing art.
You are correct. Hollywood will simply change up a couple things and then use the assets.
However, I‘m still undecided about how I think about whether generating AI art should count as Human-generated or not. On one hand, people can spend hours if not days or week perfecting a prompt with different tools like ControlNet, different promptstyles and etc. On the other hand, somebody comes up to midjourney, asks for a picture of a dragon wearing a T-Shirt and immediately gets an image that looks pretty decent. It’s probably not exactly what they wanted, but close enough, right? AI gets you 90% there what you want, and the other 10% is the super-hard part that takes forever. Anyway, sorry for dumping my though process from this comment chain on here xD
Sorry, I am firmly in the camp where that isn't art. The prompt writing can be a literary work but the result isn't a work of art. You set up the environment that allowed the image to exist but you didnt make the image.
I remember this artist who used a jet engine to throw paint onto a big canvas. Was the resulting artwork made by the jet engine, based on what you're saying?
I'm not confrontational. I just like the discussion. This whole topic is, well, fascinating.
Lol good point. Also, like when the artist uses a bucket of paint hanging from the ceiling and lets it drip on the canvas in a rotating pattern.
But the treatment of photographs in the decision fits your description. The photographer sets up the environment that allowed the image to exist but it's the camera that makes the image. The judge held that was protectable because the image represents the human's mental conception of the scene. It's not a ridiculous stretch to consider AI to be merely a camera for the prompt-writer's mental conception. I am certain this argument has been or will be tried in court. The IP owner industry is far from done litigating this topic.
Man, if AI gets to reproduce pictures exactly as I imagine them, then that's an excellent point. It's my creation. The AI just plasmated it in a screen.
But for any other scenario... it's tricky business.
Don’t apologize! It’s a really active topic and I’m super interested in other‘s opinions on this!
That latter case likely wont be copyrightable, but the former can start to meet this criteria mentioned in the article:
The way I read that, the more instruction you give to the composition of the image (ie, how detailed and descriptive you are with your prompt) the better claim you would have to copyrighting the resultant work.
I think the mistake lots of people are making is that all AI generated art is the same and should all be treated the same. Which is likely not going to be the case. And Copyright rulings are mostly done on a case by case bases, unless there is significant change this will likely still hold true and so one ruling on some AI generated art might not result in the same ruling for a different piece created in a different way with different effort.
What this case shot down is the claim that AI can claim copyright on a works as an AI is not human and copyright only applies to humans. Which is the same stance courts have tend before with content created by animals.
Great explanation, thanks!
It is if you don't say it's AI generated or you lie about how much human input it required which would be impossible to prove false.
Not impossible. If you generate something with AI and claim you created it yourself you can easily be asked to reproduce a similar works again. If you don't have the skills to do so then that is fairly big evidence that you don't hold the copy right over it. If you do have the skills, then you are far less likely to purely lean on AI generated works without putting in some more creative stances on those works, even if you are using AI as part of creating those works.
If you say you did use AI you should be able to show how much effort you are putting into creating the images, how you write your prompts, how you correct mistakes etc. All that is a skill you need to learn and it should not be so hard to show someone you do have that skill or not.
Are these definitive? No, not much evidence is definitive, but a collection of various things can help paint a picture. So there are ways to you can show if someone is likely to be lying about how much effort they put into some work. Which makes it distinctly easier than impossible to prove their claims false or not.
Asked by whom exactly? The Copyright Office? Are they going to ask for prove from every artist that requests registration for a work?
Or you can lie in your request. From the Compendium of U.S. Copyright Office Practices:
"As a general rule, the U.S. Copyright Office accepts the facts stated in the registration materials, unless they are contradicted by information provided elsewhere in the registration materials or in the Office’s records."
You can copyright the input into the machine possibly (the specific prompt). Just like we can copyright other kinds of software.
that seems quite likely.
Plenty of things that take weeks of work aren’t art.
Definitely. Ask anybody who finally evacuated the last thing they ate from Olive Garden!
Yeah while this suit covers a very specific scenario, a large majority of AI driven content does have human interaction and does qualify for copyright.
Even just a draft, fed into an AI finishing system, has some human interaction. Nothing is going to stop the AI revolution.
It looks like the key in the ruling here was that the AI created the work without the participation of a human artist. Thaler tried to let his AI, "The Creativity Machine" register the copyright, and then claim that he owned it under the work for hire clause.
The case was ridiculous, to be honest. It was clearly designed as an attempt to give corporations building these AI's the copyrights to the work they generate from stealing the work of thousands of human artists. What's clever here is that they were also trying to sideline the human operators of AI prompts. If the AI, and not the human prompting it, owns the copyright, then the company that owns that AI owns the copyright - even if the human operator doesn't work for them.
You can see how open this interpretation would be to abuse by corporate owners of AI, and why Thaler brought the case, which was clearly designed to set a precedent that would allow any media company with an AI to cut out human content creators entirely.
The ruling is excellent, and I'm glad Judge Howell saw the nuances and the long term effects of her decision. I was particularly happy to see this part:
This protects a wide swath of artists who are doing incredible AI assisted work, without granting media companies a stranglehold on the output of the new technology.
I wonder could you interpret this as AI created movie script isn't copyrightable but the actual filmed movie is. That would invite some weird competition, like we've seen over the years with the copycat movies.
This first but I don't think that is how it can be interpreted like that. Looks like it comes down to how much human input was used to guide the AI in the works. The more the human guided the AI the more they have a claim to the copyright is how I read that. Not just all AI content cannot be copyrighted. Which IMO seems like a fair way to apply copyright to AI generated content.
The latter part is basically already handled - look at any film created from a public domain works, Shakespeare plays being a big example here. I would expect non-collectable AI works adapted to film to to be handled the same way. Though I suspect that to create any good movie script with AI you would need significant human input which could lean towards to script having a stronger copyright claim by those that guided the AI.
Yes , but let's say Marvel writes the next Avengers movie with AI. Somebody else could come along and make their version of it. They'd need their own characters though, because those are copyrighted by marvel comics.
Has this been litigated yet?
Wait, he didn't just try to claim copyright over AI created material... he tried to claim the AI could copyright it?
Lol. Lmao, even.
Hope: AI gets so good that people using a personal computer can produce full TV series with a single prompt, delare it uncopyrightable, and share the best results online as a alternative to corporate stuff.
Fear: IP law becomes so disconnected from the current situation that it prompts governments start over from scratch. New IP law is written by the corporations for the corporations, and any form of creativity is restricted and monetized.
It sorta already happens. No one owns the copywrite to vampires hence all the stuff that is only slightly above fanfics becoming big. You are welcome to hate on Twilight but you can't deny it's popularity.
This doesn't change much because of a simple difference: This was an AI product put in wholesale.
There was no human intervention in (visually) creating this product, thus no human can claim copyright.
Studios aren't gonna do this when replacing some of their writers, because AI may not be good enough yet. Instead, it'll be a smaller team, they'll do the edits, and they can claim copyright.
This only really matters If AI advances to the point where we can completely create a full movie or TV show from scratch with just purely prompting, which, currently, we can not.
A spoof of Seinfeld runs 24/7 from only AI input after the initial prompt. It is bad, but exists. Depending on your quality standards, we are there. https://en.m.wikipedia.org/wiki/Nothing,_Forever
The rigging and models are copyrightable.
Having tried to do something similar, "Nothing, Forever" must have some pretty serious coding to engineer the prompts and reconstruct tiny snippets of AI generated dialogue into a full meaningful script. I wonder if that's enough for the creators to claim copyright.
That's what I see likely. Writers will eventually use this as a tool. Say setup a scene and generate 40 versions. Pick the best one, edit it, feed back what you improved, generate another 40...
We are still reading Shakespeare and I think part of the reason was how he wrote those plays.
Write the scene, pass to actors, actors have notes, rewrite scene, pass to actors, have actors act out scenes, make changes, run changes by actors, rewrite scene, perform scene, watch audiences (are they laughing at the jokes? Are they sad when they are supposed to be sad?), make edits,.....
He did it like a collaborative activity and gradual refinement. Almost none of his plays have an official indisputable version, instead we have multiple versions with slight differences.
Stop worrying about your jobs writers, this is a tool. Use it.
It's only a tool for lazy, shot by, hack writers. Like the writers for Big Bang or any kids show by Disney.
Why would a real writer need AI to steal words from others? Do you think Neil Gaimon needs help from a shitty program to write? Or any other high caliber writer? No. Because AI is for untalented artists and writers. They are the ones getting terrified that it might replace them.
Which speaks to their skillsets.
The thing is there was uncountable amout of people intervention. AI art is derivative work achieved via mathematical means.
Not in the copyright sense.
Yes, there were millions of people's work that was in the training data that was used to make whatever AI program created those AI images, but (at least right now) that isn't considered for legal ownership.
The US Copyright Office is taking the stance that there must be human effort that can be seen/pointed to in the final product directly in order to count as an "Author".
Think about that guy with the monkey taking a photo, and how that got into the public domain. Just because the company selling the camera "created the camera used to take the photo" (made the AI model) or because someone using the camera "set their own settings for the photo to be its best quality" (typed in a proper prompt for the model) doesn't mean that either party "owns" that image.
That whole paradigm could maybe change if/when AI LLM programs get seriously regulated, but even so, I personally don't think that changes the chain of ownership, nor should it.
So it's the same as when the film industry got started?
The element of human creativity derives from the Constitution.
"To promote art the author has the exclusive right." Something like that.
By that logic, all human art is derivative work achieved via biological means. No artist works in a vacuum. Everything an artist sees subtly influences their style.
This is why "intelectual property" is such bullshit
The work itself must be a product of creativity, reduced to a tangible medium.
The code that makes the art, the prompts, could be copyrighted. But not the output.
couldn't help but lol at this quote:
"To the times Disney pays people off enough..."
Oh it does adapt the time the copyright is valid.
This doesn't mean artists or movie studios can't make AI creations and sell them. It just means they can't stop people from copying and distributing them.
If a well regarded artist uses generative AI to make art, then prints a single copy or a limited edition and signs them, they can sell them. Other people can copy it, but it won't be the same. They won't have the same value as the ones the artist produced, and they won't be signed.
The hollywood model is based on ownership of IP. Can you imagine if "Stranger things" was AI generated by Netflix, had a hit first season, then Disney released a second season with new actors? Meanwhile, CBS premieres "Stranger things : Miami?"
It would be a mess and put their entire business model into a tailspin.
This ruling may be the biggest bouy the writers have gotten so far in their strike.
I think that in this scenario, Netflix could hold copyright over the idea and characters. Only the script would be out of reach. Lawyers would ensure that they hold onto the right bits to prevent this scenario.
If I asked AI to write a story for a child, the whole thing is up for grabs. If I give it characters with specific traits and a story arc, that would still be mine. Only what the AI filled in wouldn't be protected.
And I'm sure that the government would grant copyright on the human-generated inputs to an imitative large language model so-called "AI." Not sure it would be worth anything, though.
Hell, I would bet that one might be able to copyright the database that was fed to an LLM, as long as it was independently generated & created by a human and not just a hoovering of a bunch of other authors' works.
The courts have this right, for sure. Presumably we can't copyright the answer that comes out of a calculator when we hit the "=" button. But we can copyright all the formula manipulation and original thought that went into deciding which keys to press on the calculator, and possibly even the action of pressing the keys? Not sure on that last bit.
The "creation" is algorithmic, and just like the calculator's output that cannot be copyrighted. That's based on "facts" of the universe, not "creation."
Is saying "i want a long form show about 80s teenagers in a small town, one of which has psychic powers, with an overarching dark force that opposes them" really going to be "creative" enough to protect a tv series worth of output?
I think that falls apart in the same way that setting up a security camera once and then walking away doesnt give you permanent copyright over whatever it captures. There isn't enough humanity in the creation to count it as "uniquely human." The court seems to agree.
Seems to work fine with Sherlock Holmes
Holmes is very episodic by nature, which lends itself well to this structure. Even the ones that aim for an overaching story lean heavily into the "mystery of the week" for story structure.
It probally would work for things like the above, but can you honestly see long form shows working in the same way?
People could pay the studios that made the version of the show they wanted most. Instead of having no choice but to buy from whoever paid the most in the collectable copyright trading card game.
Maybe I want the "Tron 3" Dreamworks would make, instead of Disney.
I believe this is wrong. They can't copyright an AI-generated script, but the performance and film based on a script is copyrightable.
Think about this: can you copy and sell the Leonardo and Claire Danes Romeo and Juliet because Shakespeare's work is in the public domain? No. You cannot.
Disney has spent a lot of time and money making sure that their characters can't be used by anyone else, including extending copyright law each time the 'steamboat willie' version of Mickey Mouse would enter public domain.
I think they will care a lot that they can't copyright ai-generated characters.
Nfts all over again
Can they copy the artist's ai generated art including the signature and sell that?
artist's or ai generated?
If they can't copyright it, everyone can just copy their shit and it's not considered theft. I could buy that single, unique print, copy it and sell it myself more than once and make more than the dude who generated it. And it would be legal since it's public domain and not copyrighted. Would it be the same as the original? No. Do most people care about that? Also no.
ask richard stallman
Lol is that the NFT guy?
He's the emacs, GNU, GPL and copyleft guy.
Actually, it has nothing to do with human creators at all. It means that AI can't hold a copyright. But the person who wrote the article would have to actually be able to comprehend court documents to understand that, so here we are.
From the opinion:
The irony is palpable.
Hollywood will just do what they always do. Pour billions more into lobbying the government until they pass something that will allow certain exemptions.
Sonny Bono 2.0
As a person who creates both visual arts and music, though admittedly for my own enjoyment alone, I can't bring myself to ever recognize any of the AI generated stuff as Art. None of it is any good if you look at it close. It's wrong in every way. The machines were supposed to come for our jobs, but that was supposed to mean factory production and construction and shit.
It's not about being technically good or not for me, it's a question of expression. A human can express internal thoughts and feelings. An AI, at least the ones we currently have, can only do an awkward imitation. There's no intention or awareness.
I think it can be art in the same way as photography: In both cases, the human influence is far less intentional than things which start with a blank canvas, and the ease of creation means that most examples aren't art, but there are a few where someone happened to use the fullest understanding of their technical skill to capture a moment and a sensation of value. I wouldn't say all photography is art, but I wouldn't say that no photography is art, and I think generative images are similar.
I support the idea of making it uncopywrightable. I think it is obviously dependent on so many creators that granting sole use to anyone seems inappropriate.
Oh, I don't even really care about the copyright thing. I just hate for a fucking robot to lay around in it's pajamas drawing pictures while I trade 13 hours of every day to a factory for the privilege of sleeing under a roof with some food for my family and to get to lay around drawing pictures like an hour of my week. This is defined distopia.
I've still got hope. I see a sharply growing awareness of what you're point out, and I think even the billionaires are a little spooked right now.
As they should be! There's a lot more of us, and we're coming for their power!
Hmm, I think there can be a huge amount of intentionality behind photography. It's really not about the representation, it's about all the choices made. AI can represent a scene perfectly and still have no intentionality. Of course, at the extreme that gets us into thorny issues like solipsism. How can I know that anyone besides me has intentionality? Maybe everyone else is just a meat machine with no awareness at all. Or maybe everything at a certain complexity has intentionality...
The intentionality is provided as a prompt by the human author.
Yeah, that's fair. It's hard to pinpoint what feels lacking with it, but it does feel lacking somehow to me. I guess for me there's probably a tipping point where it's no longer human enough. Like, just telling an AI to make a candy forest isn't enough. But that's a straw man argument in a way. Of course someone could put a huge amount of effort into getting an AI to render exactly what they're imagining. In the end, it could be seen as just another medium. I have no doubt people are going to find incredible ways of utilizing it.
Honestly... no. In practice it doesn't work like that because while messing about and getting the AI got generate what you want you look at tons of adjacent stuff the AI comes up with which then influences what you want to see. And I bet that's a thing that even the
4k nude stunning woman with (large breasts:1.6)faction experiences, it's practically impossible to not enter a dialogue with the tool.I've done a bit of image synthesis, and I think the notion that it's without intentionality is a bit of a myth.
I wanted an image of a gorilla dressed in a polo shirt and khakis, so I prompted stable diffusion to generate some gorillas in a variety of poses, then drew a shirt on, which looked like an MS Paint drawing. Then, I ran it back through Stable Diffusion to make the crude shirt look photo realistic. I then cut the gorilla out and used it in a photo collage.
I'm not using this example to claim that I've performed art or demonstrated any skill, but the final image is definitely the intentional result of trying to take a very specific image in my head and put it on the screen.
I made a similar point in response to someone else in the thread. I agree. It's a very interesting situation to ponder. In some ways it's just another medium. The intentionality is in the people trying to produce what they're imagining via the AI. I will be curious to see what sorts of things people come up with over time.
Unfortunately at worst the machine will only improve to the point where it is unnoticable.
Its a program designed exactly to be bullied into place by humans, were just only halfway through the bullying and still coorp's are pushing it like its done.
Eventually it'll have to be accepted as just another tool by artists.
That being said I support copyright less than I do "AI rights" so I'd say this is an overall win
Fuck Corporate America and all it's Bootlicking sellout enablers.
If you think art is at all dependent on each individual recognizing it as good, then I think you've completely missed the point of a lot of art. Most art only appeals to some.
When I make art I make it for myself. I have no interest whatever if it anyone else feels one way about it. I talked with the robots. I didn't like it. Felt dirty, cheap.
As someone who creates a variety of art in pretty much any medium I find AI art to be something deep and subconscious in us. It taps into something we can't reach unless on mind altering drugs. I find it to be an amazing study of the human psyche. I have never been able to connect to most "art" especially any of the classical stuff, as well as most music. While people enjoy and even request certain pieces of art from me, all but sculpting leaves me disconnected from my work because what I visualize and what create aren't thr same. AI art has that missing piece.
So, as someone who doesn't do visual art, mainly writing and music as hobbies, my opinion is if there is intent, ie from the prompt or there editorial process of tweaking the model, then there is at last an attempt by a human to convey a message through the piece.
Whether or not it has good composition, or achieves something that resonates with a human viewer is valid criticism, but I think irrelevant as to whether or not it is art or a piece of creative expression. If someone has bad technique and can't really get their idea across well in a painting, is it no longer art? Is a painting made during a paint and sip where you're coached through the painting not art because there is no intent? These are more to gauge what you mean by art than as gotcha questions.
I would disagree that a prompt is sufficient to express human intent, specifically in writing. How many stories can you list that don't follow the heroes journey vs the ones that do? The most important part of any writing is not the setting and overarching narrative, it's the small choices the author made all along the way that make it truly human. AI can parrot those choices, but a human can't get an AI to make truly new unique decisions with any amount of prompting.
On the subject of tweaking the model, that's not really how AI models work. Users don't edit the model and keep the prompt the same to try to get different outputs. The only interface exposed to users is the input.
Author can define those small choices as part of the prompt.
I'm talking about how the story is written. You literally can't define all of those choices in a prompt, it's a continuous series of many many choices all throughout the story
Many/most of those small choices have no artistic value - it doesn't matter if you choose to use "because" or "since" for example.
Providing critical artistic choices while letting AI to make the rest of simple dumb choices (like the example above) is IMHO still creating art.
Most of what people generate with AI is shit because the people using it have no idea about art. When an actual artist picks up those tools you get quite different results.
The neat thing is that it doesn't really matter what kind of artist you are. You don't need to be a painter, a sculptor has just as trained eyes yet can prompt an oil painting. Heck I'd bet random musicians get significantly better results than the general population.
True! If u think, that something machine made could be Art should think about what art differentiate from kitsch.
If you had spent just a few minutes thinking about it, you’d have realised email jobs and creative jobs would be first on the automation chopping block.
A secretary used to schedule events and write invitations manually, now you have calendly and ChatGPT.
You used to need an HR professional to onboard new employees, now you can use on-demand courses through any of the million LMSs that exist.
Oh and there are already AI generative tools for those kinds of online courses.
Meanwhile, even the best robots move like geriatrics. We’re 100 years away from a robot that could do all that a construction worker does.
I work for a company that is automating the onboarding process even more. You type in the name of the person being on boarded and it creates the software licenses, training programs, usernames, passwords, everything, including checklists, reminders, deadlines and recurring events and renewals, with AI language models picking whether the new hire is in marketing or sales or supply chain etc
seems a pretty easy solution here. just say you've written it.. no reliable software exists for proving text is AI generated.
Doing that under oath is a crime, so those claiming copyright and their employees would be taking a lot of risk of eventually being discovered.
"Need to get away with murder? Easy! Just say 'it wasn't me, it must of been someone else who done it'. For bonus points, combine with a wink to the hot judge/juror of your choice."
Shaggy defense. It's a real thing now: https://en.m.wikipedia.org/wiki/Shaggy_defense
Do you have to add "zoinks" at the end when you say it wasn't me?
It's not like that's ever stopped any of them before.
I changed some words. Now it's mine. I dunno, error, search and replace "Mike" to "Michael".
You didn't create it so why the hell should you be able to copyright it?
Talk about an inaccurate headline. The conclusion here isn't that AI art can't be copyrighted, it's that AI cannot be a copyright holder. But it's AI, so we can't actually expect anyone to pull their head out of their ass and give it enough thought to write an article that isn't garbage.
Instead we have yet another thread about this case in which no one actually has any idea what the ruling was. Very informative.
From the opinion;
AI being the copyright holder was never even in question. Some guy tried to register AI art in his company's name, using the AI as the author of a work for hire. The Court found that he couldn't get the copyright as a work for hire since no copyright existed in the first place.
What if a monkey made art, would that be copyrightable?
No. That is in the article.
This limitation is too easy to get around. Have AI generate a picture. Take a photo of that picture and destroy the original. Copyright is now owned by the photographer. Have an AI write some music, change one note of that music and call in your arrangement of that piece, destroy the original music, and only your arrangement that you have a copyright on exists. etc.
As a photographer that knows copyright law, I assure you flat-art copying a work of art does not make it yours.
The thing I'm seeing that does sort of skirt the issue is that it's very obvious a lot of YouTubers have jumped on AI image generation to produce static images instead of drawing the images themselves or farming it out to an artist on Fiverr or something. So if they want "evil Jerome Powell with flames in his eyes" they hand it to the AI, it spits something out, and into the video it goes, to be published on YouTube as a memey splash image in the video.
Now that it's in the video, along with all the other clear acts of human creativity that form a video, it's sort of "washed" in the money laundering sense, and I don't see how you legally separate that image from the video in a way that makes the image ineligible for copyright. I don't see a court being flummoxed by that, at all. If you filch the image from the original video, or try to pull excerpts from the video featuring Evil JPow, you're in violation of copyright, and we're on pretty solid, well established legal ground with that. At the very least, you are not completely in the clear to just yank that image for yourself.
So while the original raw image of Evil Jpow that the AI spit out was not eligible for copyright by itself, now it is as part of a larger work, open and shut.
Near the end of the article it affirms pretty much that, saying, "An application for a work created with the help of AI can support a copyright claim if a human “selected or arranged” it in a “sufficiently creative way that the resulting work constitutes an original work of authorship,” [as quoted from the copyright office]
My quote is a bit messy there (i'm quoting the article who is quoting the copyright office) but you get the point.
The raw AI output, assuming no human was involved, cannot be copyrighted, but as soon as the AI output is somehow arranged into a larger work by a human, that changes everything.
So yeah, a bit of arranging, some editing, and the completely AI generated footage can be copyrighted all day. At the very least there would be a court case there.
"evil Jerome Powell with flames in his eyes" Maverick of Wall Street? lmao
You own the copyright of your photo, AI flat art has no copyright, therefore the only copyright is yours.
When you copy a public domain work, you can copyright your original contributions.
So, for example, if you create a picture book of hamlet, you'd own the layout of the text on the page, but not the text of hamlet itself.
Interesting that he filed the original copyright application as a work for hire situation. I guarantee he didn’t pay AI anything .
Yea, I spoke to Al. The dudes pissed, didn't get paid shit
So, if I use AI, then tweak it in photoshop, is us human created and now able to be copyrighted?
Basically it's how already works with cameras. Subjects and composition are enough to copyright images. Even without post-processing
The Judge is right. AI is not a living citizen.
Good.
This is the best summary I could come up with:
More than 100 days into the writers strike, fears have kept mounting over the possibility of studios deploying generative artificial intelligence to completely pen scripts.
The ruling was delivered in an order turning down Stephen Thaler’s bid challenging the government’s position refusing to register works made by AI.
Copyright law has “never stretched so far” to “protect works generated by new forms of technology operating absent any guiding human hand,” U.S. District Judge Beryl Howell found.
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.
While cameras generated a mechanical reproduction of a scene, she explained that it does so only after a human develops a “mental conception” of the photo, which is a product of decisions like where the subject stands, arrangements and lighting, among other choices.
In another case, the a federal appeals court said that a photo captured by a monkey can’t be granted a copyright since animals don’t qualify for protection, though the suit was decided on other grounds.
The original article contains 858 words, the summary contains 190 words. Saved 78%. I'm a bot and I'm open source!
Well, of course not, because since some diffusion generation are deterministic, that would mean that a specific set of parameters is now copyrighted, so nobody else gets to type in that particular set of numbers into the UI without paying the copyright holder, which of course makes no sense.
Same reason you can't copyright, say, cooking recipe for a burger.
Food and flavors aren't copyrightable or patentable because of an explicit exclusion of them. It has nothing to do with "determinism".No. Recipes are not copyrightable because they're largely functional things for instructing a process to create a food, which simply is not in the purview of copyright. Specific recipes could very well be patented, depending on the specifics. There are no "explicit exclusions" here.
I did some research, and you're right. I guess I was mislead years ago when I "learned" this.
And still the list of ingredients and food preparation process will not be copyrighted, just the way the specific recipe is written. Anyone could write a simple rephrased version of that recipe which creates the same dish and sell it. Or sell the dish in their restaurant.
You are generalizing and using the word "some" at the same time.
Yeah, because generation from "a" samplers are not deterministic I think... was trying to find the best way to word that.
Hey, I get to be lazy in my internet commenting too, OK?
Ancestral samplers are deterministic btw, but I think because they build off of the previous step it's more obvious when the determinism is broken by optimizations.
Interesting. My A1111 SD has been broken after I tried running XL so I haven't touched it in a while. I'll go test it sometimes after I reset it.
The only way to bring non-determinism into a computer is to collect physical noise, when you set a seed the computer generates a sequence of numbers (and thus noise) that is statistically indistinguishable from physical noise but actually deterministic. The ancestral samples just add more of that deterministic noise to things.
Of course the whole thing is rather moot because there's arguments to be had that physics itself is deterministic. Use of physical noise in computers is pretty much limited to situations where you want to make it impossible to guess which seed was used and thus reconstruct the noise, or, differently put: In cryptography.
To make this tangible: Remember good ole SNES games and their "push start" screens? They'd compare when you pressed the button to when the system booted up, then use that value as a seed for all the randomness in the game. As a human it's practically impossible to hit a particular seed but a computer playing an emulated SNES game can abuse such shenanigans, that's relevant in the automated speed run community.
Give them time to lobby and get a friendlier judge and try again.
If a human uses AI to create a story and then edits and tells it in greater detail the new work would be copyrighted via the editor's transformative act. This ruling is not a win for the WGA or other creatives that don't want to use AI.
AI will just become our ghost writer, which is a practice as old as writing.
Some AI work is created with a simple prompt, but the best stuff uses lots of in painting and out painting and adjustments. That would be copyrightable, there's lots of human control beyond a prompt, there are lots of cool videos of the process on YouTube. By the time AI will be able to generate full processes, videos, storylines, voices, etc, without human intervention, we won't need the movie studios either. I'm hoping we reach that in 3 years.
What's stopping someone from generating an AI script and then saying they wrote it and copyrighting it?
That would be a fraud on the copyright office. Nothing is "stopping" people from committing fraud other then it's fraud and has legal repercussions if found out.
Like if AI generated it, who is going to complain if a human copyrights it?
This is what's not computing in my head.
When they try to enforce the copyright against an alleged infringer, the infringer may claim the holder committed a fraud on the office and there is in fact no enforceable copyright. No one really knows how this would pan out in litigation because it's untread territory.
other humans perhabs
So say I create an AI that can generate movie scripts. I use it to create a script. I put my name on it and copyright it. How would anyone else know?
I'm not trying to be like, argumentative. I'm just trying to understand.
You don't "copyright something". You have a copyright on everything you create yourself by default and you don't on things that are not copyrightable. You can not put a copyright on something not copyrightable.
In practice this means if someone else copies your script without your consent, you can then try to enforce your exclusive copyright by suing them for copyright infringement. Then you need to proof and convince the judge of the originality of the work and that you put in significant creative effort.
Ok. So, it seems as usual I have missed the point here. Thanks for clarifying. So the point I'm to take away from this is that AI can't hold copyright over things they create? If the answer is "yes", then I ask. Can a human hold copyright over something an AI creates?
I am not a lawyer but the first question is probably a yes. AI is software and software does and can not hold any rights by itself as of today. The second question is what this post is about and the judge in the article said no, a human or company does not hold copyright over something AI creates. That does not mean, that anything touched by or created with the use of AI is not copyrightable. If you have your movie script error checked or rephrased with an AI tool it's still your movie script with your orignal ideas in it.
Should have figured this would be the judgment when that monkey took a selfie of itself and the ruling was that it was public domain because a human didn't take the shot, and monkies don't have rights.
I tried to reach out to Allen Iverson for a comment, but he just kept ranting about practice.
But what did Ja say? Anyone get a hold of him?
"Yo, it's Ja Rule in the house! You know, AI-generated art is an interesting twist in the game. But when it comes to copyright, I gotta say, it's a bit of a gray area. I mean, can a machine really claim ownership, right? But then again, if a human's involved in training that AI, there might be some room to argue. At the end of the day, it's all about respecting the craft and the creators, whether they're human or machine. We'll see where the courts take us on this one!" - JaRuleGPT
What is the point? Just claim it was your idea. There is no way to prove either way (AI generated or human generated). Laws need to adapt to changes.
The point is that ML can generate works in bulk outpacing humans by a ridiculous rate. And that it explicitly is meant to cover expression, but ML models don't actually express anything, they just emit statistical averages of the input.
You don't want a company with obviously way too few human employees to have created all of their works to be able to go look for similar art from others and threaten copyright lawsuits. By forcing humans to be involved in the process of creation you strongly limit the ability of such legal trolls to hurt other creators. Such copyright trolls ALREADY exists prior to ML, but extending copyright to unsupervised ML would superpower their lawsuits. They just have to spam various websites with some samples and then pretend everybody copied them.
This precedence is just what it should be. The reference to photos is completely correct. The creation of a sequence of bits isn't in itself protected, it's the selection of inputs in which the creativity lies that then carries over into protection of the output. Photos can be copyrighted because a human express something in their choice of motive. A surveillance camera for example don't automatically give its operator copyright!
And it still doesn't prevent you from using ML in the creation of things and claiming copyright, it just requires you to be the one directing the process instead of leaving it unsupervised.
Don't want to be that guy, but they absolutely don't "emit the statistical averages of the inputs". Otherwise they would create a single, most likely unicolor, image
Ok then, weighted averages.
Neither. Going from a prompt to an image is a stochastic non-linear transformation based on billions of parameters.
Your brain also performs stochastic non-linear transformations of inputs. Just in a different way.
Do I have to explain with math how my high level abstract reply applies?
Most generative ML rely on probabilities. The averages are over multidimensional complex data structures representing patterns extracted from the inputs. Like average faces when you prompt it for a face (try training it on different sets of faces and look at how the output differs, you really do see it retain averages of the patterns in the input such as average skin color and haircuts). I wasn't talking about linear arithmetic averages.
My comment is that they simply are not averages, that's it.
As a simpler example, it is like saying that a polynomial plus some noise is an average... It's simply not.
The stochastic and non linear parts are the reason these models create original images, unless overfitted.
If it was a weighted average you'd have identical, smoothed, most likely non sensical images for identical prompts.
And this is not the case.
That's all my comment.
You still misunderstand my use of "average". I am once again not talking simple averages over simple arithmetic numbers.
Look at the outputs of models trained on majority white faces vs diverse faces. If you still don't understand what I mean by averages then I guess this conversation is hopeless
Yes there's noise in the process.
But that noise is applied in very specific ways, it still fundamentally tries to output what the training algorithm indicated you would expect from it given the prompt, staying in a general neighborhood preserving the syntax / structure / patterns in the input training data related to the keywords in your prompt. Ask for a face without more details and you get a face looking like the average face in the input, usually white in most models, western conventional haircuts, etc, because that's representative of its inputs, an average over the extracted structure in the inputs. The noise just tweaks some selection of representative features and their exact properties. It is still close enough to average that I feel it is fair to call it average, because it so rarely output extremes (other than when the model just breaks down and produce nonsense).
That's a bias, not an average. Similar to human biases. Models' biases are derived from humans' biases in the training data.
Humans have biases for a male doctor and female nurse, models learn that bias unless someone intervene to identify and remove the cultural (very human) bias from the training data set
You misunderstood again. The model isn't creating the bias when it is trained on biased data. It just gives a representative output of its input. The average of many outputs will resemble the average of its inputs.
If it was a linear transformation, probably, because you'd remove the stochastic term. But transformation is non linear. I 'd be surprised if true. Do you have a reference for a statistically meaningful experiment on this?
Won't stop being that guy
Recognition;
https://odsc.medium.com/the-impact-of-racial-bias-in-facial-recognition-software-36f37113604c
https://venturebeat.com/ai/training-ai-algorithms-on-mostly-smiling-faces-reduces-accuracy-and-introduces-bias-according-to-research/
Generative denoisers and colorization;
https://www.theverge.com/21298762/face-depixelizer-ai-machine-learning-tool-pulse-stylegan-obama-bias
With generative models already used in stuff like adverts and soon in Hollywood, it becomes more relevant as it affects representation;
https://towardsdatascience.com/empowering-fairness-recognizing-and-addressing-bias-in-generative-models-1723ce3973aa
This extends to text as the output more frequently copies a style which is common in the input.
In practical terms? If you are going to generate content using AI either don't say it was AI generated or lie about how much human involvement it had. Also you can't use "this work was completely made by AI" as a hook.
So does a prompt not count as human input Edit: ok so if i train a style lora based on my own style and then prompt the ai to generate artwork , then I still don't deserve the copyright? What if I do all that and then do touchups by hand is that somehow different? I find all this stuff so silly tbh but it is interesting to discuss.
No.
If I'm sitting on the couch and I want sushi, I can open up a website, pick exactly what I want, even maybe make a few substitutions for me specificity, and get it delivered right to my house, but that doesn't mean I made sushi. I just HAVE sushi.
Anyone who has ever actually supported a real artist and commissioned work understands that they don't own the copyright, unless extra agreements have been made to transfer it. It still belongs to the original artist.
And as stated, AI can't own that. So no one does. Who would want to? It's garbled, derivative work and anyone with access to the same prompt and models could generate it themselves, which is why I find the prompt guarding so hilarious. It's all so blatantly dumb and transparent.
If you hire someone and tell them "paint my room orange", did you paint the room orange?
Wouldn't that logic give the camera people and editors the copyrights in film?
Funny that. In fact, they should totally have credit, but our copyright laws are so weighted for the property owners (rather than creators) that crediting has to be negotiated in a contract, and Hollywood lawyers are infamous for using loopholes and confusing language to remove credit and payment obligations.
To be fair copyright is fucked up, and you don't have to create something to own its rights. For example, most of the music you listen to isn't owned by the artists but the labels, and the artists only get paid if they're super popular, like Madonna or Metallica or Ye. (Which means if you pirate something as an alternative to paying for it you're denying profits to the labels, but not failing to support the artists. If you want to support an artist, go to a concert and buy one of their official t-shirts.)
So I don't think this ruling is going to slow things down. The bare minimum Hollywood studios have to do is pass AI art / copy / whatever through an editor who changes a small portion, and lo! I human made this! Then it'll go back to the courts, and they'll see it's ridiculous. (Mark Isham -- I think that's the guy -- has been tinkering with procedural music for eons, so there's counter precedent.)
Weirdly enough, it might.
I mean, sure, an orange room doesn't qualify for copyright.
But if you instruct helpers to help you make a copyrightable work to your specifications (e.g. architects instructing workers to build the building), then you own the copyright.
It all comes down to who had what part of the creative process.
For example, if I create a digital artwork and I use an AI tool to increase the resolution, I totally still own the artwork.
I am also pretty sure I retain the copyright if I let an AI fix my spelling in a story I wrote.
But if my input is negligable ("Write me a short story."), I definitely don't have the copyright.
Copyright law is complicated and there's never a clear general answer.
"For example, if I create a digital artwork and I use an AI tool to increase the resolution, I totally still own the artwork."
There are tools that allow you to convert a stick figure drawing plus a prompt into whatever you want. I can draw a stick figure holding a circle and prompt it as a bunny holding an apple and I could get the desired output. Of course thats more than increasing resolution but where do we draw the line.
That's what I said, yeah.
The line is probably drawn where it is drawn when using any other tool: For your contribution to be copyright-worthy it needs to be above the required threshold of originality. Copyright doesn't distinguish between different tools you use, it just requires you to have put enough creativity into it for it to be a creative work.
AI in this respect is very close to photography.
Just pulling out your phone and snapping a picture of something does not clear the theshold, so you have no copyright over it, no matter how detailled the picture is.
If you set up the scene and put a lot of creative input into preparation and post-processing of the picture, you will clear that threshold and you will have copyright over it.
If this sounds arbitrary and hard to pinpoint whether that threshold (which is not clearly defined anywhere) is cleared, you are totally right. This is a big issue with the copyright system, but it's also nothing that will be fixed by some people commenting on the internet.
So the question is "Did the human input into the AI tool clear the thresold or not?"
This question is the same, regardless of what tools are used, and it's a common part of lawsuits concerning copyright. There is no blanket answer and it needs to be checked on a case-by-case basis. That's how copyright law works.
we let the AI draw it for us
On one hand, great; will that extent to software development, architecture and other fields?
On the other hand, sounds like the first step to, when AI and androids reach self awareness and conscience, legally keep them enslaved.
When an AI can make that argument for themselves, then the law can change, until then, a human must be part of the creative process to hold copyright.
A classic example is the monkey selfie. There's no copyright because there was no human involved in the creation of the selfie.
Well, the opressed and enslaved usually has no say about changing the law.
It's a good thing that AI isn't capable of being oppressed or enslaved. Because it's currently less AI and more, janky code that does a thing, and sometimes does it correctly.
For now. 40 years ago, what it does now was impossible science fiction.
And for the next 40 years it will likely remain science fiction.
So there's no point in fucking up all the case law for something that doesn't exist. Seriously, copyright needs to be cut down, not expanded further. It's already the life of the author plus 70 years. How does that even work? Copyright is meant to get humans to produce more creative works, so how the fuck does that work after death?
The answer is, corporations that don't die. They want more control, and want AI to make shit, so they don't have to pay real people to do it.
So no. No copyright for theoretical AI. no copyright for monkeys with names assigned by some third party. Just stop trying to expand copyright.
My 2c:
On projects over a certain revenue the AI could say how much it was influenced (trained on) by the respective copyrighted content and then royalties could go out to the people who own that content in percents.
My 4c:
There could be an intellectual property blockchain and everything that can be used to train an AI gets a token.
Again, I think all of this should only be mandatory for huge corporations, similar to how unreal engine is free under 1 million dollar earnings.
This could also be an interesting way to see how human made content makes its way through the "minds" of AIs.
No.
Copywrite law doesn't distinguish based on the size of the corporation, and I don't think it should.
That would probably be picocents per artist. How do you want to transfer that amount of money?
So basically anything generated procedurally just lost its copyright protection because of this wacky definition of "mental conception." Anyone with skill and talent can exert control over the output of an AI. Likewise, someone with no idea how to paint will not produce what they're "mentally conceiving."
That may be true if the LLMs are wholly-deterministic. Speaking of procedural generation, humans don't take its outputs as-is and call it a product. As it is, it's more like running an RNG while you specify the seed number.
So if I use a hammer to create art I can't protect it because I used a hammer? Exchange hammer for AI
From my reading it was more like if you created art with a hammer the copyright doesn't automatically go to the hammer which would then transfer to you via the "work for hire" clause. So if you then say lent out that hammer to a bunch of other artists to make art with you would theoretically have a copyright claim to everything they made using your hammer.
Only if it is some sort auto-magic hammer that you can give a description of what you want, but the hammer actually decides what the final piece of art looks like and produces it on the medium without you needing to make the strikes.
Now if you took the output of that auto-magic hammer and than significantly reworked it to meet your vision, that derivative output would be copyrightable.
Giving a description of what you want doesn’t count as producing it is what this seems to be saying. Which is not in favor of people who just want to plug random things into a black box to use that black box as a cash machine.
This also means that the person who owns the black box can’t steal ownership of your output by merely being the owner of the black box. This would be equivalent to you leasing a printer and the company you’re leasing it from saying they own anything you print.
So good ruling. You can copyright the output if YOU rework the output, but you can’t copyright the direct output.
Way I see it a hammer is a tool, like a paintbrush or a camera or Photoshop or chatGPT.
If you use the hammer to break a plate and call it art, you get the copyright.
If you set the hammer up on a machine and feed it a million plates to smash, but with your direction and intent - to choose the types of plates, speed of the hammer, to use the tools to output different results - its still art and you still get the copyright.
But if your hammer sits inside a Platesmasher 9000 which randomly takes plates as input, selects which plates to smash, smashes them, assesses the results, smashes more, then outputs a perfect smashed plate without you doing anything - that's not copyright able. You can't sya you deserve the copyright, as you did not meaningfully contribute to the work - the Platesmasher did everything. You cant point to the output of the system and say "the system made that and deserves copyright" because it's just an algorithm, it doesn't know or have intent behind what it's doing, and can't be assigned a right.
What this does is stop corporations from building a million Platesmashers and claiming copyright on a billion versions of smashed plates - human intervention is required as part of the creative process to use the tools in order for there be a right in the first place.
None of you AI boosters can make a coherent and cogent argument in good faith. Why is that?
Maybe you could get an AI to write a better argument. The hammer doesn't steal others work to make its own.
You could train a LLM using only your own art, and then generate art that is based solely on your own. Do you not own that AI generated art?
The way I understand it if you did that and tried to take just the raw AI output and get a copyright on it you couldn't based on this ruling. But if it was one of the tools you used to create a piece of art even if it was just editing and making small changes to it to suit your creative vision based on what the AI put out then you could. It sounds like the judge is mainly talking about works solely generated by AI.
Also my understanding was that this guy was trying to get the AI generator itself to be considered the author for the things it generated for the purposes of copyright. Which would theoretically transfer to whatever entity is running the AI because of the "Work for Hire" clause.
Did you steal other's work to learn how to write that sentence?
It's a good thing they weren't making an argument then - but asking a (flawed) question. Just like comparing machine learning to stealing is a flawed comparison.
It's a bit of a flawed comparison (AI vs a hammer) - but let me try.
If you put a single nail into wood with a hammer, which anyone with a hammer can also do, and even a hammer swinging machine could do without human input, you can't protect it.
If you put nails into wood with the hammer so that it shows a face, you can protect it. But you would still not be protecting the process of the single nail (even though the nail face is made up of repeating that process many times), you would specifically be protecting the identity of the face made of nails as your human artistic expression.
To bring it back to AI, if the AI can do it without sufficient input from a human author (eg. only a simple prompt, no post processing, no compositing, etc) it's likely not going to be protectable, since anyone can take the same AI model, use the same prompt, and get the same or very similar result as you did (which would be the equivalent of putting a single nail into the wood).
Take the output, modify it, refine it, composite it, and you're creating the hammer equivalent of a nail face. The end result was only possible because of your human input, and that means it can be protected.