AI Created images (generated by Midjourney) lose registered copyrights

[Kisai]

 

 

Summary

 AI Created images (generated by Midjourney) lose registered copyrights.

 

Quotes

Quote

AI-created images lose U.S. copyrights in test for new technology

 

Feb 22 (Reuters) - Images in a graphic novel that were created using the artificial-intelligence system Midjourney should not have been granted copyright protection, the U.S. Copyright Office said in a letter seen by Reuters.

"Zarya of the Dawn" author Kris Kashtanova is entitled to a copyright for the parts of the book Kashtanova wrote and arranged, but not for the images produced by Midjourney, the office said in its letter, dated Tuesday.

 

Quote

The Office has completed its review of the Work’s original registration application and deposit copy, as well as the relevant correspondence in the administrative record.1 We conclude that Ms. Kashtanova is the author of the Work’s text as well as the selection, coordination, and arrangement of the Work’s written and visual elements. That authorship is protected by copyright. However, as discussed below, the images in the Work that were generated by the Midjourney technology are not the product of human authorship. Because the current registration for the Work does not disclaim its Midjourney-generated content, we intend to cancel the original certificate issued to Ms. Kashtanova and issue a new one covering only the expressive material that she created.

My thoughts

 

Well, so using Midjourney to create an image leads to uncopyrightable images.

 

We're probably going to see a few more of these stories for similar "full disclosure" of AI generated materials (eg GPT3/ChatGPT) but there will be further consequences for uses of AI in accessibility spaces (eg, someone who used an AI to write a story and generate images, because they're unable to use their hands) but other that that, I feel that there's going to be some deceptive copyright fights where fights to invalidate a copyright registration will be on the basis that the copyrighted work shows no WIP/draft versions to assure human authorship.

 

 

Sources

https://www.reuters.com/legal/ai-created-images-lose-us-copyrights-test-new-technology-2023-02-22/

https://fingfx.thomsonreuters.com/gfx/legaldocs/klpygnkyrpg/AI COPYRIGHT decision.pdf

 

Update:

 

So the guidance can be summarized as "you must disclose the use of AI and the prompt used"

 

Guidance https://www.govinfo.gov/content/pkg/FR-2023-03-16/pdf/2023-05321.pdf

https://arstechnica.com/tech-policy/2023/03/us-issues-guidance-on-copyrighting-ai-assisted-artwork/

[Holmes108]

Things are gonna get messy.

[jaslion]

7 minutes ago, Kisai said:

AI in accessibility spaces

For creative arts I highly doubt it.

 

Good that these things are losing copyright.

[LAwLz]

46 minutes ago, jaslion said:

Good that these things are losing copyright.

Why is it good that AI generated art can't be copyrighted? 

[jaslion]

11 minutes ago, LAwLz said:

Why is it good that AI generated art can't be copyrighted? 

 

Decent video on it

[Fasterthannothing]

24 minutes ago, LAwLz said:

Why is it good that AI generated art can't be copyrighted? 

I agree it's dumb and honestly I'm probably gonna be flamed for this but I don't see a big deal with AI art being able to be copywritten. If AI is really generating unique pieces every time then it's going to be unique no matter what.

 

[05032-Mendicant-Bias]

Yes, no copyright for non humans (or non corporations, because corporations clearly are people, right? )

No copyright for animals either. PETA argued a monkey should own the copyright for the selfie they made.PETA lost. That picture has no copyright beause animals can't have copyright. (of course corporations can own copyright, because they clearly are people, right? )
https://en.wikipedia.org/wiki/Monkey_selfie_copyright_dispute

 

Computers can't own copyright, IMO is good so long as computers aren't sentient. When it's starting to become really hard to argue that an AGI isn't sentient, perhaps as a civilization we will need to revisit this position. Or turn off the power switch. Whichever works better for long term survival.

 

4 minutes ago, Fasterthannothing said:

I agree it's dumb and honestly I'm probably gonna be flamed for this but I don't see a big deal with AI art being able to be copywritten. If AI is really generating unique pieces every time then it's going to be unique no matter what.

There needs to be a minimum amount of human touch for a work copyrightable. Pressing Generate on Stable Diffusion doesn't pass that treshold. Probably applying a gradient filter is enough. For the human handler to make a copyrightable work.

[jaslion]

23 minutes ago, Fasterthannothing said:

I agree it's dumb and honestly I'm probably gonna be flamed for this but I don't see a big deal with AI art being able to be copywritten. If AI is really generating unique pieces every time then it's going to be unique no matter what.

 

Thing is it's not because it has repeatedly committed plagiarism over and over including showing signatures of others

[porina]

The law is what the law is.

 

Personally I don't get why assigned copyright isn't possible from a non-human to a human. At some point a human will be responsible, so why can't that person hold copyright from machine or other non-human generated content?

 

For example, if a human took a photograph of the AI generated art, would the photo be copyrightable? Or would it require some level of changes to create a derivative work? I don't know the details there of how it is defined and/or interpreted.

[goodtofufriday]

6 minutes ago, porina said:

The law is what the law is.

 

Personally I don't get why assigned copyright isn't possible from a non-human to a human. At some point a human will be responsible, so why can't that person hold copyright from machine or other non-human generated content?

 

For example, if a human took a photograph of the AI generated art, would the photo be copyrightable? Or would it require some level of changes to create a derivative work? I don't know the details there of how it is defined and/or interpreted.

From what I understand: if there is framing and composition to the photograph take then that can be copyrightable. but if you took a photo so close up that just the image but basically in a larger frame, it wouldnt be copyrightable.
If the image in the photo in question has a copyright then i believe that photo becomes the copyright of the image rights holder, not the photo taker. 

[LAwLz]

33 minutes ago, jaslion said:

 

Decent video on it

That video does not address my question.

Why do you believe that it is a good thing AI generated art can not be copyrighted?

 

I'd prefer to get an answer from you without having to watch a 12 minute video full of half-truths, misinformation and arguments based on emotions if possible.

[jaslion]

Just now, LAwLz said:

That video does not address my question.

Why do you believe that it is a good thing AI generated art can not be copyrighted?

 

I'd prefer to get an answer from you without having to watch a 12 minute video full of half-truths, misinformation and arguments based on emotions if possible.

The lack of human input.

 

Art has always been an expression of us humans. Once you remove that it becomes soulless and just sludge in an ever increase pile of other sludge.

 

That is my core reason for wanting AI art gone.

 

I don't dislike ai stuff out right it can be a great assistive tool potentially but instead of that it's meant to replace us humans.

 

 

I am personally feeling the effects of this as basically 50% of my inquiries are hey I have this ai generated thing and it's not working or exactly how we want can you fix it for us quickly/alter it/...

It's not quickly it's basically redoing it all again as it's one big blob of crap usually that is unworkable.

 

This will change as the more it gets art from the internet and the models improve the less mistakes happen and eventually it will mean that there will be hundreds of thousands of artist accounts that post a new art piece ever couple of minutes before it all implodes in on itself. Leaving the human artists behind.

 

All we will soon have is physical media which is a whole lot less accessible.

[LAwLz]

44 minutes ago, 05032-Mendicant-Bias said:

There needs to be a minimum amount of human touch for a work copyrightable. Pressing Generate on Stable Diffusion doesn't pass that treshold. Probably applying a gradient filter is enough. For the human handler to make a copyrightable work.

Where is the line?

Is a human creating the prompt and tweaking settings in the program enough work?

What if it gets edited in Photoshop afterwards before publishing?

What if the human also created the model used?

What if a human assembled all the images used for the dataset to create the model?

What if a single human created all the images in the dataset as well as tweaked the settings?

What if a human made the prompt, and combined multiple models as well as settings such as applying LoRAs, tweaking the weights, maybe did some in-painting using AI and so on?

 

I feel like there is a line that can be drawn somewhere, but it's highly subjective and I feel like a lot of people are applying separate criteria and reasoning whenever something that contains the words "AI" gets brought up.

[BabaGanuche]

The problem with having AI getting copyright is problems like this, https://www.vice.com/en/article/wxepzw/musicians-algorithmically-generate-every-possible-melody-release-them-to-public-domain, where the AI exhaustively copyrights a particular area.

[LAwLz]

1 minute ago, jaslion said:

The lack of human input.

How do you define "human input"? 

I think some AI generated art pieces have a ton of human input into them. Far more than a lot of other computer generated content.

 

 

2 minutes ago, jaslion said:

Art has always been an expression of us humans. Once you remove that it becomes soulless and just sludge in an ever increase pile of other sludge.

What is to say AI generated art can't be an expression for a human? 

If I have a clear idea in my mind, and creates an image that matches that using an AI art program, is that not an expression from me?

 

 

3 minutes ago, jaslion said:

That is my core reason for wanting AI art gone.

But this thread is not about AI art going away...

I feel like you read the title, saw that it contained the words "AI" and then went on an anti-AI rant that has nothing to do with the thread itself.

 

So I'll ask you again. Why do you want Ai art images to not be copyrightable? In what way does that improve the world? Remember, we are talking about AI art images being copyrightable or not, we are not talking about the ethics of AI generated images, we are not talking about your feelings about AI generated art images or anything like that. We are just talking about if AI generated art should or should not be copyrightable and what consequences those particular decisions would have.

[LAwLz]

4 minutes ago, BabaGanuche said:

The problem with having AI getting copyright is problems like this, https://www.vice.com/en/article/wxepzw/musicians-algorithmically-generate-every-possible-melody-release-them-to-public-domain, where the AI exhaustively copyrights a particular area.

But that's not really related to AI. You could do the same without AI as well, which that article demonstrates.

But luckily (or unluckily depending on how you look at it) it is very easy to just look at that and do what the judges did, and throw it out on a case by case basis.

 

People would not be able to generate a ton of AI art images and then claim copyright on all of them because as that article proves, it has already been tried without the involvement of AI and it was thrown out.

 

Saying that AI would open the floodgate for such exploits is completely unfounded. There is zero evidence to support that idea. In fact, we have evidence that disproves it, such as that article you linked.

[05032-Mendicant-Bias]

19 minutes ago, LAwLz said:

Where is the line?

Drawn arbitrarly based on the interpratation of some judge in each jurisdiction.

 

Personally I think the whole copyright system is archaic and needs to be redone from scratch anyway. How is it fair that a work is protected 70 years after the death of the author, and meanwhile IP is juggled between corporations like expensive trading cards?

 

I think it's fair that only humans can own copyright. I would upgrade it to "only sentient can own copyright" as it would pretty much cover all cases and move the conversation on the intent and spark of the creator (sorry maquaque, you still don't get a copyright in my book).

 

AI Generators require a prompt, as long as your prompt has a "minimum effort", I think it its output should quilify the prompter for a copyright. 

[jaslion]

20 minutes ago, LAwLz said:

So I'll ask you again. Why do you want Ai art images to not be copyrightable? In what way does that improve the world?

Gets the money out of them. Which means it's a less desirable now to overtake this space. Which means the space stays open for humans to enjoy without big corporations pushing everyone out.

[Thaldor]

And I still question how using AI is any different from using apprentices or, the more modern term, assistants to create your art (that you sign and is presented as made by you and you are the sole copyright owner)?

 

From the point of copyright and morality how is it any different that someone used AI and takes the credit for themselves than, for example, Kehinde Wiley taking credit over a painting made by one of his Beijing hub workers or Jeff Koons taking credit from his (up to 150) studio assistants in New York?

 

If you say it's about the "human touch" then what is the threshold? If one person makes a sketch and another makes the piece from said sketch is the artist the one who made the sketch or the one who made the piece? Is that any different from making a sketch or just describing the piece of art to an AI who then creates it?

What I have understood you cannot copyright an art piece made by AI for yourself because it is not made by human but apparently it's completely fine to copyright an art piece made by someone else as yours as long as they have been in contract with you (or just worked around you like in the case of Andy Warhol and his Factory, whatever someone created in the Factory, it was piece by Andy Warhol even if his input to it was just owning the space where it was made). We are completely fine for "artists" stealing the credits as long as it's based on at least some kind of vague contract but doing the same for a software which cannot even get sad about the fact is completely wrong.

 

Not to even go to the lengths of what is art and what can be copyrighted. Like some big name artist taping a banana to a wall is art but some random dude doing the same is just a banana taped to a wall and if done later the "artist" can sue the random dude for taping a banana to a wall for copyright infringement because why exactly? If the random person did the thing earlier and tried to copyright it, they probably would be laughed off because taping a banana to a wall isn't art, but everything turns upside-down once the random person has done some art and gained name as an artists in the past and suddenly the exact same thing becomes art, so is the art the banana taped to a wall or the name of the person who taped a banana to a wall? And we cannot miss the point here, would it be any different if the "artists" said to a person to tape a banana to a wall and then claimed the copyright for this magnificent piece of art that he did or that the "artists" programmed a robot to tape the banana to the wall and it wouldn't be copyrightable art because the artists didn't steal the copyright from a person.

[goodtofufriday]

My opnion overall summed up by Miyazaki when AI generated art/animation was pitched to him.Im a IT admin in corporate world and I see nothing good coming from AI for civilzation in general. Humans have had a hard enough time adapting socially to the explosion in technology in the past 30 years. We arent ready for AI. 

 

[Kisai]

48 minutes ago, LAwLz said:

Where is the line?

Asking the wrong question. That should be "what is the minimum threshold of human input required?

 

Which is answered.

 

Quote

Rather than a tool that Ms. Kashtanova controlled and guided to reach her desired image, Midjourney generates images in an unpredictable way. Accordingly, Midjourney users are not the “authors” for copyright purposes of the images the technology generates. As the Supreme Court has explained, the “author” of a copyrighted work is the one “who has actually formed the picture,” the one who acts as “the inventive or master mind.” Burrow-Giles, 111 U.S. at 61. A person who provides text prompts to Midjourney does not “actually form” the generated images and is not the “master mind” behind them. Instead, as explained above, Midjourney begins the image generation process with a field of visual “noise,” which is refined based on tokens created from user prompts that relate to Midjourney’s training database. The information in the prompt may “influence” generated image, but prompt text does not dictate a specific result. See Prompts, MIDJOURNEY, https://docs.midjourney.com/docs/prompts (explaining that short text prompts cause “each word [to have] a more powerful influence” and that images including in a prompt may “influence the style and content of the finished result”). Because of the significant distance between what a user may direct Midjourney to create and the visual material Midjourney actually produces, Midjourney users lack sufficient control over generated images to be treated as the “master mind” behind them.

48 minutes ago, LAwLz said:

Is a human creating the prompt and tweaking settings in the program enough work?

Nope.

48 minutes ago, LAwLz said:

What if it gets edited in Photoshop afterwards before publishing?

Also nope.

Quote

Finally, Ms. Kashtanova suggests that she personally edited some of the images created by Midjourney. Her letter points to two specific images contained in the Work. While the Office accepts the statement that the changes were made directly by Ms. Kashtanova, it cannot definitively conclude that the editing alterations are sufficiently creative to be entitled to copyright. First, Ms. Kashtanova explains that she “modif[ied] the rendering of Zarya’s lips and mouth” in an image on page 2 of the Work. Kashtanova Letter at 12.

The changes to Zarya’s mouth, particularly her upper lip, are too minor and imperceptible to supply the necessary creativity for copyright protection. The Office will register works that contain otherwise unprotectable material that has been edited, modified, or otherwise revised by a human author, but only if the new work contains a “sufficient amount of original authorship” to itself qualify for copyright protection. COMPENDIUM (THIRD) § 313.6(D). Ms. Kashtanova’s changes to this image fall short of this standard. Contra Eden Toys, Inc. v. Florelee Undergarment Co., 697 F.2d 27, 34–35 (2d Cir. 1982) (revised drawing of Paddington Bear qualified as a derivative work based on the changed proportions of the character’s hat, the elimination of individualized fingers and toes, and the overall smoothing of lines that gave the drawing a “different, cleaner ‘look’”).

 

48 minutes ago, LAwLz said:

What if the human also created the model used?

Likely it would have be curated and entirely from their own previous works, but that's not what the copyright office cares about.

48 minutes ago, LAwLz said:

What if a human assembled all the images used for the dataset to create the model?

Same as above. If you assembled a specific corpus to create a specific model to give a specific result, that would likely be "sufficient control over the input", but that's still something of a stretch, since no single human can produce enough artwork to train an art model. There are 171,000 words in the english language, and the average artist could produce maybe 1 work per week at an acceptable level needed to train an AI, so assuming someone had 60 years of work to train a model on, at 1 per week, They would have only produced enough for 0.2% of the training data. No rather, you would need to get 1000 artists who can already do the same house style, and the only business that will have that kind of access would be Disney.

 

48 minutes ago, LAwLz said:

What if a single human created all the images in the dataset as well as tweaked the settings?

See above, a single human can not create a big enough dataset for an AI to learn from. You would need at the minimum, 1 visual symbol per english word, and that english word has to appear every time it's used in every work.

So "monkeys playing poker", "dogs playing poker" and "dogs playing ball" need to all have the same symbols in them that appear in no other work. Your typical piece of art is just not going to label every specific detail.

 

48 minutes ago, LAwLz said:

What if a human made the prompt, and combined multiple models as well as settings such as applying LoRAs, tweaking the weights, maybe did some in-painting using AI and so on?

As per the copyright office, the image itself has to be created by the human. Period. It doesn't matter about anything after the fact.

 

48 minutes ago, LAwLz said:

I feel like there is a line that can be drawn somewhere, but it's highly subjective and I feel like a lot of people are applying separate criteria and reasoning whenever something that contains the words "AI" gets brought up.

 

Just based on this copyright office ruling, the line is literately:

As the Supreme Court has explained, the “author” of a copyrighted work is the one “who has actually formed the picture,” the one who acts as “the inventive or master mind.”

 

Thus if you start an image by an AI prompt (let's ignore img2img for this argument) alone, it can not be copyrighted, as it's the human who has formed the picture in their mind that does, and that is not something that the AI is capable of doing.

 

A more interesting test would be if img2img style transfer is subject to the same. As this is the technology the artists are actually afraid of.

[goodtofufriday]

4 minutes ago, Thaldor said:

And I still question how using AI is any different from using apprentices or, the more modern term, assistants to create your art (that you sign and is presented as made by you and you are the sole copyright owner)?

 

From the point of copyright and morality how is it any different that someone used AI and takes the credit for themselves than, for example, Kehinde Wiley taking credit over a painting made by one of his Beijing hub workers or Jeff Koons taking credit from his (up to 150) studio assistants in New York?

 

If you say it's about the "human touch" then what is the threshold? If one person makes a sketch and another makes the piece from said sketch is the artist the one who made the sketch or the one who made the piece? Is that any different from making a sketch or just describing the piece of art to an AI who then creates it?

What I have understood you cannot copyright an art piece made by AI for yourself because it is not made by human but apparently it's completely fine to copyright an art piece made by someone else as yours as long as they have been in contract with you (or just worked around you like in the case of Andy Warhol and his Factory, whatever someone created in the Factory, it was piece by Andy Warhol even if his input to it was just owning the space where it was made). We are completely fine for "artists" stealing the credits as long as it's based on at least some kind of vague contract but doing the same for a software which cannot even get sad about the fact is completely wrong.

 

Not to even go to the lengths of what is art and what can be copyrighted. Like some big name artist taping a banana to a wall is art but some random dude doing the same is just a banana taped to a wall and if done later the "artist" can sue the random dude for taping a banana to a wall for copyright infringement because why exactly? If the random person did the thing earlier and tried to copyright it, they probably would be laughed off because taping a banana to a wall isn't art, but everything turns upside-down once the random person has done some art and gained name as an artists in the past and suddenly the exact same thing becomes art, so is the art the banana taped to a wall or the name of the person who taped a banana to a wall? And we cannot miss the point here, would it be any different if the "artists" said to a person to tape a banana to a wall and then claimed the copyright for this magnificent piece of art that he did or that the "artists" programmed a robot to tape the banana to the wall and it wouldn't be copyrightable art because the artists didn't steal the copyright from a person.

I think your arguement generally fails because its humans interacting with humans with humans doing all the interpreating. 

If we had the food making machines from star trek, would you call someone who told a machine "make a michelin 5 star meal" a chef? Probably not. So why is someone who tells a machine "make this piece of art" an artist? 

It all about humans hands or mind physically interacting with the space around us, taking bits and pecies of things to create something new. A machine doing the bits and pieces part is not the human creating. 

[Holmes108]

1 hour ago, LAwLz said:

But that's not really related to AI. You could do the same without AI as well, which that article demonstrates.

But luckily (or unluckily depending on how you look at it) it is very easy to just look at that and do what the judges did, and throw it out on a case by case basis.

 

People would not be able to generate a ton of AI art images and then claim copyright on all of them because as that article proves, it has already been tried without the involvement of AI and it was thrown out.

 

Saying that AI would open the floodgate for such exploits is completely unfounded. There is zero evidence to support that idea. In fact, we have evidence that disproves it, such as that article you linked.

 

Agreed. A lot of opinions seem based on either emotions (understandably to a degree) or based on opinions on current copyright laws, and/or what they should or shouldn't be. 

 

But I'd rather ignore all of that. We can grant some obvious truths, like the current system for humans is probably imperfect, and that AI is also often guilty of infringement. But all of that is fixable, in theory. So no point in saying "AI shouldn't be able to copyright because they steal". That's only going to get better and better to the point of that it's "inspiration" will be indistinguishable from human inspriation.

 

So what about the broader principle then? Assuming you agree with any sort of copyright laws in general (and fair enough if you don't), there can be an argument made for it to cover AI. For example, it seems logical to me, for it to simple protect originality, period. Meaning if an AI comes up with an original work (which YES, I believe it absolutely can), then it should just be covered. The alternative is that another artist could then paint the exact same picture, copying the AI, and profit from it, which also doesn't feel like it's in the spirit of the rules.  

 

It's a complicated issue, to be sure, but I definitely think there's a very reasonable argument to be made.

[Kinda Bottlenecked]

Seems like a benefit to society if AI art isn't copyrightable. Which means it can't be taken down or claimed either I assume.

[BabaGanuche]

1 minute ago, goodtofufriday said:

If we had the food making machines from star trek, would you call someone who told a machine "make a michelin 5 star meal" a chef? Probably not. So why is someone who tells a machine "make this piece of art" an artist? 

To elaborate on this you would you call someone commissions a painting the artist of the painting? No that would be absurd. If I give the same request to multiple painters, you would get different results. So the painter is the artist. I give the same request to current day "AI" and everyone who give the same request gets the same results. The "AI" is not creative.