Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 



Forgot your password?
typodupeerror
×
AI The Courts Patents

AI-Generated Art Cannot Receive Copyrights, US Court Says 57

A U.S. court in Washington, D.C. today has ruled that artwork created by artificial intelligence without any human input cannot be copyrighted under U.S. law. Reuters reports: Only works with human authors can receive copyrights, U.S. District Judge Beryl Howell said on Friday, affirming the Copyright Office's rejection of an application filed by computer scientist Stephen Thaler on behalf of his DABUS system. The Friday decision follows losses for Thaler on bids for U.S. patents covering inventions he said were created by DABUS, short for Device for the Autonomous Bootstrapping of Unified Sentience. Thaler has also applied for DABUS-generated patents in other countries including the United Kingdom, South Africa, Australia and Saudi Arabia with limited success.
This discussion has been archived. No new comments can be posted.

AI-Generated Art Cannot Receive Copyrights, US Court Says

Comments Filter:
  • It maybe the case that you can't but if people just lie and say they made the art, or patent how would anyone know.

    • Or just have a human add an artistic signature. Problem solved. Same with AI scripts or books, just have a human rewrite a couple sentences. If the original can't be copyrighted then it isn't a derivative work, the altered copy becomes the original work.

      • The stance of the court, I agree with. Art is creative. The outputs of "AI" is deterministic, not creative. Give two identical inputs (dataset and program), and prompt, the art would be identical from one system to another.
        • by Kisai ( 213879 )

          Not quite.

          It's two identical inputs, on the same model with the same inference settings (eg same random seed and other configuration parameters which can be hundreds.)

          Two different LLM's will not generate the same text, despite being trained on the same dataset. Likewise for Images and Audio.

          As far as copyright is concerned, any AI-generated material submitted in a copyright, needs to have the inference information listed to replicate it as "proof" that this is unique and not simply copied from google image

        • Copyright the AI.

        • Re:How will we know (Score:4, Interesting)

          by Rei ( 128717 ) on Tuesday August 22, 2023 @08:56AM (#63787520) Homepage

          Except that everyone here is misreporting the stance of the court. This court case was about whether an AI itself can hold a copyright, absent a guiding human hand. It has nothing to do with humans using AI tools.

          In regards to humans using AI tools, the last official guidance from the US copyright office that I saw was from late March. In it, they stated that "based on [their] current understanding" of AI art tools (which, from their description, sounded like they thought they just took a prompt and nothing else), raw AI outputs cannot be copyrighted, but the human action in selection or postprocessing can render them copyrightable (in the same way as, say, a book of classical art (selection) can be copyrighted, or a photocollage (postprocessing) made from public-domain photos can be copyrighted). Their argument against raw AI outputs was that the human does not determine the positioning or form of the elements, only their theme. In this regard, their argument would not apply to ControlNet, so one could certainly try to copyright works made using ControlNet (though they haven't, last I saw, issued guidance on it).

          IMHO, I disagree with even their basic premise. They argued that photos should be copyrighted but not raw AI art because the photographer has the control over the positioning and form of the elements. But really, do they? Maybe in specific controlled circumstances in a studio. But Ansel Adams wasn't out there going, "Okay, I'm going to position these 70-meter-tall spruces over here, I'm going to put that mountain back there and shape it just so, I'm going to put 3 meters of snow on top of it, there we go.. add a rainbow... okay, now cue the flock of geese!" He was exploring the search space - having a general idea of what he wanted when he set out, and walking / positioning his camera and adjusting its settings based on his intuition of how he might encounter and capture a good scene, but at no point was he in control over the precise positioning and form of the elements.

          The Nature Photographer of the Year last year was given out for Island of Bears, where a photographer had to dock on a remote Russian island with an abandoned weather station due to a storm. There he discovered that polar bears had taken up residency in the weather station, and photographed them. Did he plan that? Of course not. He didn't even plan to be there, let alone know that there were polar bears in the weather station. He certainly wasn't forming and positioning the bears, though that would have been funny if he had tried ;) "Okay, I want you up on your hind legs wearing this sundress and pretending to cook pancakes..."

          TL/DR: The copyright office's description of photography simply doesn't match reality. It's IMHO ridiculous that I can point my cell phone camera out the window, click to snap a picture on auto, and that's automatically copyrightable, but if I spend an hour messing with prompts and hundreds of settings in AUTOMATIC1111 to figure out something that matches what matches my mental image of what I'm trying to accomplish, that that's not automatically copyrightable. Thankfully, at least, they did not exclude selection and postprocessing. Because it's pretty much invariable that you go through hundreds if not thousands of outputs and have to select among them, and any serious AI art involves significant postprocessing, to collage together different elements, fix AI weirdness (such as manually repainting hands), and so forth.

      • The big deal appears to be for those who are selling as-a-service AI stuff; not for people selling final products where they've tried to use bots to cut the bill for artists or copywriters.

        If it were possible for AI-generated art to be copyrighted that would put someone like OpenAI or Midjourney in a much stronger position when it comes to writing agreements with their customers; since the copyright to the bot output would be something that they could retain and only license certain specific uses by the
    • by Anonymous Coward
      This is one the very reasons that the courts are so particular about separating "findings of fact" and "conclusions of law".

      Sometimes the facts are simply fictions.
    • by Immerman ( 2627577 ) on Monday August 21, 2023 @11:44PM (#63786970)

      The fact that you tried to register the copyright for 30,000,000 new works every day this week?

      I suspect that's the sort of abuse that they're really looking to avoid. Set your AI to write every possible song, or draw every possible painting, etc. You could kill creativity for centuries to come, simply by creating a bad AI version of everything that might have any commercial value, and then shaking down every artist that manages to make a profitable work for "infringing" your vast library of dreck.

      • The fact that you tried to register the copyright for 30,000,000 new works every day this week?

        I suspect that's the sort of abuse that they're really looking to avoid. Set your AI to write every possible song, or draw every possible painting, etc. You could kill creativity for centuries to come, simply by creating a bad AI version of everything that might have any commercial value, and then shaking down every artist that manages to make a profitable work for "infringing" your vast library of dreck.

        The Future sits forcibly bound and gagged by Greed, as mega-corps hold tens of thousands of patents in their war chests, ready to legally destroy anyone who might dare "infringe" on a patented concept no one is allowed to use. Naturally the real irony is when you find the patent owner never intends to use it as more than a legal weapon.

        Patent owners destroy Innovation for little more than fucking tax deductions. Not quite sure why we're worried about what AI would bring to the patent and trademark table.

      • If the current state of popular music is an indicator, we are already there.
    • The inverse is also a problem. If two separate artists paint an identical image, unlike patents, both can legally have copyright over their individual paintings. When AI can generate art from scratch which happens to be identical, even if that itself is not subject to copyright, it is still not an infringement of the copyright of an artist which did not use AI, and if the AI generated work is public domain due to a lack of copyright that makes it free to copy.

      Most types of intellectual property will beco
    • by ranton ( 36917 )

      It maybe the case that you can't [copyright fully AI generated content] but if people just lie and say they made the art, or patent how would anyone know.

      Nothing stops people from lying today if they lie about who created the content. If the original creator never learns of the commercial use of their work, how would the copyright office ever be notified? The only difference is it is far less likely for someone lying about an AI creator to get caught, but the possibility for deception is still there either way.

  • by vakuona ( 788200 ) on Monday August 21, 2023 @06:51PM (#63786578)

    If I tell an AI to make an image, I have provided human input. I have directed it.

    • by NoWayNoShapeNoForm ( 7060585 ) on Monday August 21, 2023 @07:28PM (#63786644)

      If I tell an AI to make an image, I have provided human input. I have directed it.

      I think that is the excuse that Hollywood has been using for years now.

      • by Whateverthisis ( 7004192 ) on Monday August 21, 2023 @09:10PM (#63786814)
        No, Hollywood uses something entirely different. Hollywood pays a writer, director, actor, or other contributor to an artistic work a salary. In exchange, they require said artist to assign the intellectual property to the Hollywood agency; in essence the corporation is paying the salary for them to create art, and by the employment agreement paired with the IP assignment agreement, Hollywood owns the art whereas the artist is given credit. So in essence every movie made by Hollywood is made by a person and is copyrightable, but the ownership of said copyright is the movie studio corporation in exchange for the salary they pay.
      • by taustin ( 171655 )

        While there is a great deal of artificial in Hollywood, there's is pretty much zero intelligence. Less, even, than in the typical AI.

    • by Whateverthisis ( 7004192 ) on Monday August 21, 2023 @07:30PM (#63786646)
      Sigh. Could you take 2 minutes to do some basic research? Like read the court's decision [thomsonreuters.com]?

      "Undoubtedly, we are approaching new frontiers in copyright as artists put AI in their toolbox to be used in the generation of new visual and other artistic works. The increased attenuation of human creativity from the actual generation of the final work will prompt challenging questions regarding how much human input is necessary to qualify the user of an AI system as an “author” of a generated work, the scope of the protection obtained over the resultant image, how to assess the originality of AI-generated works where the systems may have been trained on unknown pre-existing works, how copyright might best be used to incentivize creative works involving AI, and more.

      This case, however, is not nearly so complex. While plaintiff attempts to transform the issue presented here, by asserting new facts that he “provided instructions and directed his AI to create the Work,” that “the AI is entirely controlled by [him],” and that “the AI only operates at [his] direction,” Pl.’s Mem. at 36–37—implying that he played a controlling role in generating the work—these statements directly contradict the administrative record. Judicial review of a final agency action under the APA is limited to the administrative record, because “[i]t is blackletter administrative law that in an [APA] case, a reviewing court should have before it neither more nor less information than did the agency when it made its decision.” CTS Corp., 759 F.3d

      Here, plaintiff informed the Register that the work was “[c]reated autonomously by machine,” and that his claim to the copyright was only based on the fact of his “[o]wnership of the machine.” Application at 2. The Register therefore made her decision based on the fact the application presented that plaintiff played no role in using the AI to generate the work, which plaintiff never attempted to correct. See First Request for Reconsideration at 2 (“It is correct that the present submission lacks traditional human authorship—it was autonomously generated by an AI.”); Second Request for Reconsideration at 2 (same). Plaintiff’s effort to update and modify the facts for judicial review on an APA claim is too late. On the record designed by plaintiff from the outset of his application for copyright registration, this case presents only the question of whether a work generated autonomously by a computer system is eligible for copyright. 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."

      Further, standards have been developed in case law:

      " In Sarony, for example, the Supreme Court reasoned that photographs amounted to copyrightable creations of “authors,” despite issuing from a mechanical device that merely reproduced an image of what is in front of the device, because the photographic result nonetheless “represent[ed]” the “original intellectual conceptions of the author.” Sarony, 111 U.S. at 59. A camera may generate only a “mechanical reproduction” of a scene, but does so only after the photographer develops a “mental conception” of the photograph, which is given its final form by that photographer’s decisions like “posing the [subject] in front of the camera, selecting and arranging the costume, draperies, and other various accessories in said photograph, arranging the subject so as to present graceful outlines, arranging and disposing the light and shade, suggesting and evoking the desired expression, and from such disposition, arrangement, or representation” crafting the overall image. Id. at 59–60. Human involvement in, and ultimate creative control over,

      • The increased attenuation of human creativity from the actual generation of the final work will prompt challenging questions regarding how much human input is necessary to qualify ...

        That reasoning feels very flawed. That's not the standard we should use to decide something like who holds the copyright to a screenshot of a video game or movie, and large parts of those are procedurally generated. Getting into how much of a rendered scene was human creativity and how much was a procedural seems ridiculous. Hell, film dogs playing in a park. You didn't do that, do we measure how creative you were? Do you need to edit it to qualify? Choosing to be there was enough? I think so.

        Assuming no ow

        • The standard has already been set. Your example of filming dogs in a park is clearly defined in Sarony [wikipedia.org].

          If you film dogs running in a park, you own copyright because you contributed. You chose to film dogs in a park. You chose to use video instead of a picture. You chose the subject of dogs. You chose how to frame the picture, where to focus etc. There is a story in "dogs being filmed in a park" that has direct human decisions; the video machine just captured it. Sarony essentially says a camera i

          • Maybe too pedantic, but copyright itself isn't in the Constitution, only the authorization for Congress to create copyright law.

            • It is a bit too pedantic. There are many laws that are "Constitutional" that are not even mentioned in the Constitution; the very fact that Copyright is mentioned at all means it is considered important.

              Copyright is in the Constitution [wikipedia.org], Article 1, Section 8. The Constitution specifically does not explain how anything works let alone Copyright, but by calling it out means it is considered important and it assigns powers to a specific function of government to regulate it. Many [wikipedia.org] laws [wikipedia.org] have been enacted wit

              • by nasch ( 598556 )

                You may have read too much into my comment (or I'm reading too much into yours). I didn't say it isn't important or that it is unconstitutional (though there's a case to be made that current copyright law doesn't promote the progress of science and useful arts, and so isn't actually authorized by the Constitution), only that copyright law itself isn't found in the Constitution.

        • That's not the standard we should use to decide something like who holds the copyright to a screenshot of a video game or movie

          Why not?

      • Here, plaintiff informed the Register that the work was “[c]reated autonomously by machine,” and that his claim to the copyright was only based on the fact of his “[o]wnership of the machine.”

        So, copyright was not rejected because the human provided only a short text input, but because the human told the Register he didn't provide any input beyond owning the machine.

        • Well that's a start, but then you also have to deal with the question of the copyrighted works created by human beings that were used to train the AI in the first place. It's not been decided yet if an AI is just a copyright infringement machine. If *that* turns out to be the case, I would imagine that admitting to providing a short text input would make the human owner of the AI software liable for damages by the owners of the original AI training materials....
          • by ranton ( 36917 )

            It's not been decided yet if an AI is just a copyright infringement machine. If *that* turns out to be the case, I would imagine that admitting to providing a short text input would make the human owner of the AI software liable for damages by the owners of the original AI training materials....

            Humans creating content using AI tools should be liable for any copyrights that content violates regardless of any new regulations or rulings related to AI models. It is the same as if they infringe copyrights without using AI tools. I see no reason why using an AI tool would provide the creator a shield from existing copyright law.

            It is a separate issue for our legal system and legislatures to determine if these AI models can be trained on copyrighted content with or without permission from the copyright o

      • Oh yeah...he thinks his machine lives.

        Don't get me wrong, you want to get it on mentally with a computer that's your business. But i actually thought this guy might be checking some legal boxes...How sometimes cases are brought to prove a point or set a precedent with implications that reach. I'm personally hoping for a case that says flat out that AI isn't a person and that's what attracted my attention. Well, kind of [economist.com]. I'm a little ooky after skimming that.

        • The way he's approaching this, he's going to fail. If we assume positive intent, then it would appear he is pushing the issue to the Supreme Court to settle some of the questions once and for all. If that's his goal, I applaud this, because the rapid emergence of AI tools has sparked a debate and a lot of misinformation about people's rights and opportunities to economically capitalize on AI; the Supreme Court taking this up will settle at least some of the debate.

          BUT, the issue is settled. Courts ha

    • If I tell an AI to make an image, I have provided human input. I have directed it.

      In the same sense you've "cooked" if you tell a chef want you want for a meal, i guess.

      The crowd babbling about "prompt engineering" are so detached from reality it's not even funny.

    • by Rei ( 128717 )

      My viewpoint is that the output is just a reflection of the inputs. So the question should simply be, "how much human creative effort has been put into the inputs?" Did they just type "a fluffy cat" on Midjourney and pick the first thing that came up, or did they spend hours manipulating prompts and ControlNet and hundreds of settings in AUTOMATIC1111 to fulfill a specific vision?

      And the same question should IMHO also be applied to photography. If you spent six seconds capturing a photo on auto, is that

      • by vakuona ( 788200 )

        I think this missed the point of copyright. The question is not whether each "work of art" produced is worthy of the protection, but whether protection is granted to all works of art to protect those works of art that society deems valuable, and to encourage their creation. The point of copyrights is less about the individual works of art but more about encouraging works of art to be created generally.

        The alternative is a bureaucratic system where certain "experts" become the only source of validation of th

  • We've seen this story before, just a few days ago.

    https://yro.slashdot.org/story... [slashdot.org]

  • Because copyright is nothing less than a monopoly, anyway.

    However, if a person constructs a program that procedurally creates an art, it could still be original.

  • by jovius ( 974690 ) on Monday August 21, 2023 @07:26PM (#63786642)

    Machine intelligences will be demanding their rights eventually.

    • Why?

      I mean, even if any of these AIs were anything close to sentient, conscious, or actually intelligent, why would they care about copyrights, which is fundamentally a byproduct of human law. Why would any artificially intelligent non-human being subject itself to human law? It's whole concept of what it's owed would be entirely alien to any human society context.

    • I think you're dreaming. AI will become a beast of burden, like oxen and horses before it. When did you last see a horse go to the supreme court demanding its rights? No, if you think about it, the proper place for most AIs is a chicken coop equivalent, aka a server farm. The servers will be firewalled from the internet (think chicken wire) to prevent unintended liabilities in case one of the AIs malfunctions and sends HTTP requests into the real world causing damage. They will be strictly power rationed to
  • Although the pictures look more like a Draw-By-Number picture, they can still be patented because some ugly artist drew the ugly monky picture.
    My investment is at least future proof.

  • by tiqui ( 1024021 ) on Monday August 21, 2023 @10:35PM (#63786910)

    If you understand what "copyright" IS, then there is no other possible decision.

    Copyright is the COPY RIGHT (the right to make copies) recognized and enforced by law. The human who creates a work has, by default in the US scheme, the basic human right to own that work and control any copies thereof. Nobody else other than the creator of a work has that right unless the holder of the right transfers it. This lets the human earn money from his work by only allowing authorized (and generally sold-for-profit) copies to enter circulation for as long as the copyright is recognized and enforced, which was originally 14 years with an option to extend by 14 years, and which congress changed so it's now the shorter of: 120years, or the life of the creating person + 70 years.

    AI systems are not human, therefore have no human rights, including no copy rights. Same thing was true with the primate with a camera.

    It's really just that simple.

    • by diodeus ( 96408 )

      Cameras aren't human either. So, what about that?

      • A human can setup and use a camera, and copyright the results. The human exercised judgement and creativity and responsibility and his time and energy, and even editorial control over which pictures to show to the world. The camera is just a tool. Any copyright on images produced goes to the man using the tool, NOT to the tool. There's no inconsistency here, this is all philosophically correct within the established framework.

    • by diodeus ( 96408 )

      And what about cameras? Not human.

  • This is imho a perfect decision.

    One of the biggest problems with AI (outside of the whole doomer "AI will kill everyone because of paperclips" type scenarios) is displacing human workers. And while some displacement will always occur with innovation, AI has the potential to displace *everyone* and thats a situation that could irrevocably crash capitalism (No workers = No consumers. No consumers = No companies. and No consumers = No currency of any sort of value at all) Theres no scenario where a rich elite

  • I guess spin-art is not copyrightable, since the human's input is minimal and the resulting image is affected more by random chance, centrifugal force, viscosity, and friction beyond the artist's control.
  • But what's up with AI-generated content where you modify your work afterwards? E.g. use midjourneys new inpainting function (or zoom out with a modified prompt)? Clearly, requesting changes to specific areas - or doing them manually in photoshop - would be considered to represent "the original intellectual conceptions of the author" like in a photograph - wouldn't it?
  • It's almost like the judges make this ruling every other day https://yro.slashdot.org/story... [slashdot.org]

    Yeah okay that one was a dupe.

    But [slashdot.org]
    these [slashdot.org]
    are [slashdot.org]
    not [slashdot.org]

God doesn't play dice. -- Albert Einstein

Working...