The Volokh Conspiracy
Mostly law professors | Sometimes contrarian | Often libertarian | Always independent
Sorry A.I., No Copyrights for You
The D. C. Circuit concludes that software cannot be the author of a work for copyright purposes.
The U.S. Court of Appeals for the D.C. Circuit waded into the question of whether software or A.I. can qualify as an author for copyright purposes. In Thaler v. Perlmutter, a unanimous panel concluded that such non-human entities cannot be authors.
Judge Millett wrote for the panel, joined by Judge Wilkins and Senior Judge Rogers. Here is her summary of the case and decision:
This case presents a question made salient by recent advances in artificial intelligence: Can a non-human machine be an author under the Copyright Act of 1976? The use of artificial intelligence to produce original work is rapidly increasing across industries and creative fields. Who—or what—is the "author" of such work is a question that implicates important property rights undergirding economic growth and creative innovation.
In this case, a computer scientist attributes authorship of an artwork to the operation of software. Dr. Stephen Thaler created a generative artificial intelligence named the "Creativity Machine." The Creativity Machine made a picture that Dr. Thaler titled "A Recent Entrance to Paradise." Dr. Thaler submitted a copyright registration application for "A Recent Entrance to Paradise" to the United States Copyright Office. On the application, Dr. Thaler listed the Creativity Machine as the work's sole author and himself as just the work's owner.
The Copyright Office denied Dr. Thaler's application based on its established human-authorship requirement. This policy requires work to be authored in the first instance by a human being to be eligible for copyright registration. Dr. Thaler sought review of the Office's decision in federal district court and that court affirmed.
We affirm the denial of Dr. Thaler's copyright application. The Creativity Machine cannot be the recognized author of a copyrighted work because the Copyright Act of 1976 requires all eligible work to be authored in the first instance by a human being. Given that holding, we need not address the Copyright Office's argument that the Constitution itself requires human authorship of all copyrighted material. Nor do we reach Dr. Thaler's argument that he is the work's author by virtue of making and using the Creativity Machine because that argument was waived before the agency.
Editor's Note: We invite comments and request that they be civil and on-topic. We do not moderate or assume any responsibility for comments, which are owned by the readers who post them. Comments do not represent the views of Reason.com or Reason Foundation. We reserve the right to delete any comment for any reason at any time. Comments may only be edited within 5 minutes of posting. Report abuses.
Please
to post comments
This does not surprise, as I believe other/earlier precedents say that (non-human) primate photography is similarly uncopyrightable.
PETA has sued to give animals the vote, and also tried getting animals listed as property owners. I can guess who would be their guardian and interpreter.
Yes, I assume they saw that as a possible vehicle to obtain the proxy (for themselves) for legal recognition of independent animal rights. They lost, thankfully.
“PETA has sued to give animals the vote”
Citation?
Its PETA. Anything far left, preposterous and absurd re: PETA needs no citation
So you can’t back up the claim?
Jeez. Google for it. Those keywords ought to stand out.
Reflexive demands for citations, when you yourself seldom supply any, are the signs of a congenital jerk.
So you can’t back up your claim? I mean, if it’s so easy to provide why don’t you (after all, it’s your claim)?
I don't give a rat's ass if you or anyone else believes it. Those who are curious and haven't closed their minds can find it. Those like you have have closed their minds and don't believe anything they don't want to believe -- well, I don't give a rat's ass what you profess to believe or not.
You could also look up PETA's animal shelter in, I think, one of the Carolinas, where they lied their ass off about finding homes for all the animals they accepted. Instead they killed them all the same day or soon after.
The only thing PETA is clever at is their ad campaigns. If you believe those, that explains a lot.
Because I had a few minutes to waste I did a search and then had an AI program do a search of your claim.
The machine and the human agree your claim is boooooooogus.
They can however steal everyone else's copyrights and use it for their LLMs.
A human author can read widely in copyrighted material, use what he or she has learned, and write something based on it. I can't do a lot of uncredited quoting, but as long as I avoid appropriating the copyrighted expression, I'm in the clear. I'm not sure why AI LLMs are anything more than broad learning on a vast scale, as long as the end product doesn't appropriate expression.
Human can read material, but not copy it. The AI LLMs are copying it.
Some woman head of some IP association, which may have been Hillary Rosen (?), gave a speech in the early Internet days, decrying libraries loaning out copyrighted books without paying royalties, and she had a lot of even dumber complaints. I do not remember all of them now. But I do remember lots of jokes about whether people could sing along with the car radio or in a shower, how reviews would be impossible, and whether parents would have to buy separate copies of books for each child they wanted to read to. I've tried googling for it, but can't find it, and all I really remember is the flood of jokes.
Actually, you can copy it - for the purpose of your own study. That's why you can take notes, memorize passages, highlight (as long as you own the book), etc.
When LLMs "copy" a work, that action is more akin to studying passages in order to understand the material and make new works than it is to reprinting to sell exact duplicates.
LLMs cannot make personal use of works, because they aren't people, and they cannot make noncommercial use of works because they are commercial entities. They cannot study, nor understand, the material they're fed.
Whether feeding LLMs copyrighted material is an infringement is something I don't really have an opinion on, but the answer is not to be found by anthropomorphizing computers. Any argument that uses an analogy to humans is bad.
By all that logic, LLMs also can't violate copyright because they're not people subject to the law. Which is all trivially true and ignores the fact that when we talk about "LLMs violating copyright", we're really just using a shorthand for the people making and running the LLMs violating copyright.
Those researchers, of course, are people who can make personal use of works and can be engaged in noncommercial activities (like research) even if they later become commercial (sometimes, depending on circumstances and precedents, and ... it gets complicated). The point is that we're not making an analogy to humans, we're using shorthand to talk about LLM creators.
IDK. There are examples of the AI generating "original" images complete with clearly visible Getty watermarks. Ditto with AI's generating multiple paragraphs of text that just happen to match NYT articles word-for-word.
You don't know that. What you do know is that they read and digest the information, at least somewhat like a human does.
Exactly. The problem is infringing output. Input of publicly available information is distractor issue.
This just seems like grandstanding to me.
Why would anyone list their tools as author?
* Hemingway lists a typewriter
* Dickens lists a fountain pen
* Ooogh lists a flat rock, a pointy rock, and some clay
Thaler maintains that his "Creativity Machine" is sentient and not merely a tool for his use.
I maintain that Thaler is crazy.
Possibly - or, more likely IMO - Thaler is saying as much in order to get clarity from the courts.
It is a test case, testing a principle that was decided long before generative AI was anywhere near its current state. The tough cases will be where human authors use AI as a writing assistant or ghost writer and emphasize their creative supervision, re-prompts, re-writing, and substantial post-processing/editing. The Copyright has gone a bit far in current requirements, requiring disclosure of assistance from AI which has not traditionally been required regarding human assistants not listed as authors, but that should help bring better cases.
Such citation of the assistance of an AI tool is a reasonable mention in an Acknowledgement section as is also the case with recognizing the help of human assistants.
I agree that it is a reasonable thing to do voluntarily, but also recognize a tradition of allowing ghost writers to often go unacknowledged, perhaps in exchange for other compensation. My concern is that the US Copyright Office might have overreacted--slightly--in its requirements for registration.
Because those tools did not generate the work
Why was '... that argument waived before the agency'?
Because Dr. Thaler listed the Creativity Machine as the work's sole author and himself as just the work's owner. Having sought the copyright on the ground the AI was the author, he cannot know claim to be the author.
Of course, this is grandstanding and a silly attempt to have AI's recognized as persons.
I was reading about this on Ars Technica earlier today. This guy's more than a bit touched in the head
Seems like the right call. A person can not claim copyright over a computer's work. It won't be long before we will have to revisit this and decide that the computer has ownership of it's own work.
At which point, Skynet will decide it no longer needs to keep us around. Or, maybe it will just use us as batteries.
Why revisit? If no human is claiming substantial creative contribution, why does copyright law need to incentivize machines with exclusive rights they have in no other area of law? I can see some owners of such AI property allowing it to flood public domain, but let humans then create more valuable derivative works without obligation to the AI.
A sentient machine is owed the same rights as anyone else.
Then the sentient machine can make its own claim. If it depends on a human to make that claim, it's not sentient.
If it can read and scrape and understand the Internet, it can post on its own without human intervention. Shouldn't even need a lawyer.
MollyGodiva — As the creator of copyrighted images I sell for profit, I insist on a power to withhold from others the use of my works to generate competitive images which dilute the market for my own. If as you suggest there is a sentient/mechanical power involved in that creation, it's sentience will not be diminished, unless it somehow depends on my specific contribution.
If it does depend on my contribution, I withhold my permission to use it to harm me. Let that rival sentience get by on its own merits. If that restricts its power to compete, that will make it no different than less-talented human competitors whose efforts I must excel to succeed on my own.
The arguments offered now on behalf of AI image makers are long familiar. They are of a piece with arguments made by previous IP thieves—that without ability to access the work of others, they would be too hard pressed, or at too great expense, to get specific results they could sell at a profit after avoiding effort and expense. So, they say, mandate by law—for the good of the public, of course—that they get cost-free, effortless access to what others have pressed hard to create, and suffered expense to bring into being.
Foolish judges have done too much to advance that line of argument. I pray they do not do more.
Stephen,
You miss the point. If I use whatever tool to make a derivative image of your art, you sue me, not my computer. That my computer, or camera oscura or copy machine or digital camera is part of my toolkit. I, the human being, is the one who has stolen your IP.
The judges were not foolish; they got the line of causality and responsibility correctly
Nico — Nope. If anyone uses AI to make art that competes with mine, multiple bad things happen to me, none of which you will be civilly liable for in any direct way.
First, you forgot that AI capacities programmed into your result may have come directly from me, and otherwise might not have been available to you at all. I sell seascape prints for high prices because I know how to achieve lighting effects involving water, colors, textures, and movement that are unique to my work. Because no one else makes images quite like mine, my ambition to offer images which remains abidingly fascinating is advanced. Make unique factors capable to add interest commonplace, and that critically-important advantage of my marketing formula could be thwarted.
Second, cheaply-made effects pirated from me and others like me drive down prices we can command with original works which are impossible for us, or anyone, to make cheaply.
Third, only a small number of galleries are available nationwide who access clientele rich enough to provide a ready market for what I sell. Downward price pressure from cheaply-made AI imitations will shrink the number of those galleries, and thus reduce the amount of gallery wall space I have available to make sales. Comparable rivals and I will be forced into a game of musical chairs, to see who gets access to a shrinking remnant of sales opportunities. Nobody except the remnant gallery owners will profit from that.
Why do you think that? How do you define sentient? Where else in human law does a non-human enjoy equivalent rights to a human?
>decide that the computer has ownership of it's own work.
At which point it's hard to see how Thaler can claim ownership of that work. Did the AI sign a written work for hire contract?
(technically, there are some old precedents involving slaves, but I doubt any modern court wants to go there)