US Appeals Court Rejects Copyrights For AI-Generated Art (yahoo.com) 33
An anonymous reader quotes a report from Reuters: A federal appeals court in Washington, D.C., on Tuesday affirmed that a work of art generated by artificial intelligence without human input cannot be copyrighted under U.S. law. The U.S. Court of Appeals for the District of Columbia Circuit agreed with the U.S. Copyright Office that an image created by Stephen Thaler's AI system "DABUS" was not entitled to copyright protection, and that only works with human authors can be copyrighted.
Tuesday's decision marks the latest attempt by U.S. officials to grapple with the copyright implications of the fast-growing generative AI industry. The Copyright Office has separately rejected artists' bids for copyrights on images generated by the AI system Midjourney. The artists argued they were entitled to copyrights for images they created with AI assistance -- unlike Thaler, who said that his "sentient" system created the image in his case independently. [...]
U.S. Circuit Judge Patricia Millett wrote for a unanimous three-judge panel on Tuesday that U.S. copyright law "requires all work to be authored in the first instance by a human being." "Because many of the Copyright Act's provisions make sense only if an author is a human being, the best reading of the Copyright Act is that human authorship is required for registration," the appeals court said.
Tuesday's decision marks the latest attempt by U.S. officials to grapple with the copyright implications of the fast-growing generative AI industry. The Copyright Office has separately rejected artists' bids for copyrights on images generated by the AI system Midjourney. The artists argued they were entitled to copyrights for images they created with AI assistance -- unlike Thaler, who said that his "sentient" system created the image in his case independently. [...]
U.S. Circuit Judge Patricia Millett wrote for a unanimous three-judge panel on Tuesday that U.S. copyright law "requires all work to be authored in the first instance by a human being." "Because many of the Copyright Act's provisions make sense only if an author is a human being, the best reading of the Copyright Act is that human authorship is required for registration," the appeals court said.
Quite right (Score:4, Insightful)
Copyrights are temporary--this is required by the Constitution.
Current law ties that to the life of the author.
The law makes no provision for granting copyright for something whose "author" can't die.
Re: Quite right (Score:2)
Not just that, but copyright is supposed to be for the purpose of protecting creative human effort. It can't do that if it's busy protecting software output that far outnumbers the human creations.
Re: (Score:2)
Authors and inventors are not merely implied to be humans, just because words are different doesn't mean they are unrelated. These laws are written in a language, a language has a vocabulary with accepted meanings, if we dismiss any meaning because it suits a narrative we have no language and no law.
I see no cars here, though I'll admit I smell the implication in "Fords and Chevys".
Re: (Score:2)
"It can't do that if it's busy protecting software output that far outnumbers the human creations."
You might want to think about that one some more.
Re: (Score:2)
Spurious argument, since copyrights can be, and are, held by corporations, which in theory, can last forever. Time limited, still, but it's simply a fixed term not connected to anybody's lifespan.
This ruling isn't in any way about duration, it's about human input into the creative process.
Re: (Score:2)
Corporations cannot be original authors though, original authors are the interests that are served. Original authors can transfer rights to corporations.
And it's not clear your point anyway, original authors' interests are not protected because original authors can die. A human lifespan, or lack thereof, is not the concern, it's the need to promote creativity. Congress doesn't need to promote corporate creativity.
Re: (Score:3)
Corporations cannot be original authors though, original authors are the interests that are served. Original authors can transfer rights to corporations.
There has to be human creativity behind the work, but corporation can, and do - often - hold the original copyright. It is called "work for hire," [wikipedia.org] and it's explicitly part of Title 17.
"A work made for hire (work for hire or WFH), in copyright law in the United States, is a work that is subject to copyright and is created by employees as part of their job or some limited types of works for which all parties agree in writing to the WFH designation. Work for hire is a statutorily defined term (17 U.S.C. 101) and so a work for hire is not created merely because parties to an agreement state that the work is a work for hire. It is an exception to the general rule that the person who actually creates a work is the legally-recognized author of that work. In the United States and certain other copyright jurisdictions, if a work is "made for hire", the employer, not the employee, is considered the legal author. In some countries, this is known as corporate authorship. The entity serving as an employer may be a corporation or other legal entity, an organization, or an individual."
In other words, you have no clue what you're talking about.
Re: (Score:1)
Any artwork commissioned by a corporation from a human as work for hire is property of the immortal corporation.
And the copyright on that artwork is good for the shorter of 95 years from publication or 120 years from creation.
Recall the constitution in Article I, Section 8, Clause 8, often called the Copyright Clause says
"To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries."
And these were
Copyright on what basis? (Score:2)
What basis did this Thayer guy claim he could claim copyright? The AI system allegedly created the "art" "spontaneously" without his input so why does he think he can claim copyright?
I also tend to doubt either word applies here, "spontaneously" or "art."
Human Input (Score:2)
This ruling doesn't seem to apply to most of the "AI" art that people are concerned with. Because those systems require human input, namely, a prompt.
The prompts are crafted by a human.
In some systems, the "AI" takes other human inputs. We don't grant Copyrights to typewriters, either. (Nor to operating systems that complain "not a typewriter".)
As for the purely "spontaneously" generated art, it seems to me that it should be owned by the human who operated the machine. (That is, the owner of the AI, which m
This is correct (Score:2)
We need less IP laws, not more
AI generated stuff may be fun, but it's not art
Modern copyright law is bs (Score:1)
While I agree that artists, musicians, authors, etc should be able to make a profit for a reasonable amount of time, what we have today is insane.
First of all, the entire point of copyright was so the creators could make a profit in a reasonable amount of time then the work goes into public domain. Society as a whole benefits this way.
Second, the musicians don't get the bulk of the money for their works. The records companies do. The same can be said of authors and book companies. So the "it protects the artists" line is absolutel bullshit.
Third, I find it amazing that a musician, writer, artist should be able to benefit from something they did decades prior for the rest of their life and their kids and grandchildren as well. Someone who flips a burger doesn't get to collect royalties for each burger they flipped for the rest of their life. I don't get royalties for any medical publications I've been a part of, or procedures I came up with. In the medical field things that save lives don't make any profit unless we create a device that takes years of testing and thousands or millions of dollars before getting approval.
That being said, why in the hell should something that a computer generated deserve a copyright for someone who owns the tool that created it. Does the company that made the brushes and paint an artist used to create a painting get to own the painting?
Cybersquatting (Score:2)
So you have AI crank out images of unique things... humans, vehicles, mascots, building architecture, every permutation of music you can come up with, and copyright it all. Then go after anything similar claiming it is a derivative work. This is just one reason why we can't have AI generated stuff become copyrighted.
Re: Cybersquatting (Score:2)
Re: (Score:2)
Maybe.
A lot of people take Jackson Pollock very seriously as an artist. But with some of his art, it sure looks like he just randomly (i.e. arbitrarily, as in your example) splattered stuff around. It's either over my head, or below my willingness to acknowledge. (I'm not here to talk shit about the guy, but I'll admit I never "got it
Re: (Score:2)
IANAL but I think doing something as small as applying a canned filter (i.e. "sunset" or "vintage b&w") probably meet the standard if human input, or adding meme text overlay. The standard exists and isn't very high, but if you didn't know it was there you'd probably to over it getting it posted online.
Re: (Score:2)
*probably trip over it
Re: (Score:2)
Is AI generated SOFTWARE copyrightable then? (Score:3)
If Software is subject to the same copyright law,
then does this mean, that AI-generated software
is also not subject to copyright?
Interesting implications, if so.
Especially considering that Microsoft allegedly trained copilot
on the Github *private* repositories as well.
Theoretically, by querying an AI trained on confidential data and code, ... one could "launder" copyrighted code,
and asking it to generate using that knowledge, software which is similar
in operation and outputs
into a non licenced but legal, form.
Ahh, the lawyers will love this.
Re: (Score:1)
Re: (Score:1)
Thus, does it mean that any software if either open source or public domain ?
No copyright claim can be issue for any generated code. This seems to implies one of two alternatives:
- either the software is open source, then the uncopyrightable part of the code can be shown, is small, and then the remaining software is copyrightable
- As most of all small binary parts of a software may be demonstrated as AI generated, any copyright claim is quite difficult to defend for the whole software.
Makes sense to me (for a change) (Score:3)
The chances that various people are going to get essentially the same image out of essentially the same AI setup approach 1. I find near-duplicates in the output of my diffusion models all the time, and they don't have the same seed. Some models just have some very noticeable preferences, perhaps because they're under-trained and don't know how to do anything else, perhaps because they're overtrained and too many tokens point at the same area in their latent space. It doesn't even have to be the same model, or even the same type of model. More than once I've shared a prompt with an SDXL user, and gotten practically identical results with a FLUX model. What are the chances someone else in the world is going to ask for "a gnome pirate girl with a budgie on her shoulder"? I'd have to imagine they're pretty high, and Flux Schnell doesn't give a whole lot of variety in its output when this is what you ask for. I really liked the result I got, but it would be exceptionally foolish to think that I somehow "own" that when it could (and probably will) pop out of someone else's rig if they ask for the same thing and let it churn for an hour or two.
Now if you choose and apply some filters, crop it, do color corrections, etc. and still someone comes up with almost exactly the same image... then at the very least, they were trying to get the same aesthetic you were. This is exactly why there has to be a human element for copyright. If you cannot in good faith say "that result could not have happened if I hadn't carefully guided the machine onto it", then you must accept that someone else is highly likely to end up in the same place. You need to be able to say "I strongly suspect this is stolen because I did X, Y, and Z in these very specific ways" or you need to shut up and take your lumps. Diffusion models repeat themselves a lot and have zero awareness that they're doing it, that's just how it works.
Seedy workaround... (Score:1)
Legal break of DRM ? (No author, no DMCA) (Score:2)
DRM law protection justification is written as protecting the works of authors. No author, no DMCA. Thus, if AI generated art content (Book, Music, Movies, ...) becomes ubiquitous (for books, it seems to be already the case. Spotify shows it may also be the case for music), it becomes lawful to break DRM and distribute both tools to break DRM and the uncopyrightable content.
What is AI expressing? (Score:2)
Is a computer program trained on mounts of data even capable of self expression? Aren't the proponents of AI copyright reaching to far in trying to anthropomorphize an algorithm?
And if AI is capable of self-expression, is it ethical or legal to own one? A bit of a pandora's box if we want to recognize AI art as something that can be copyrighted.
Photographs (Score:2)
Re: (Score:2)
On the other hand, its not at all clear that someone taking a brush in hand and copying Picasso's work would be a violation of its copyright.
Why is that not clear? Isn't it obviously a derivative work?