When Camille Roux sets out to make a work of art, he often turns to the internet for advice.

“What’s your favorite?” he recently tweeted, sharing four computer-generated variations on an abstract visual theme, along with a poll allowing people to register their preferences. The denizens of Twitter began to weigh in. Some cast a vote without comment, while others offered Roux a rationale. One user said they preferred a particular image over another “because the red makes it look more lush.” The user speculated that they might feel differently if the overall color scheme were different. In response, Roux offered up a new slate of candidates, similar to the original four, but with their dominant lines now dramatically set off against a black background. The dialog continued as Roux considered other possible permutations.

This is one artist’s approach to making “generative art.” Art historians use the term to refer to any art practice in which the artist cedes some control over the final product to a system—like a computer program or machine—that is to some degree autonomous.

Roux enhances his work’s indeterminacy by quasi-crowdsourcing certain artistic decisions, but chance is already incorporated on the level of the work’s code. He provides an input, and the system generates an image as an output. “And every output is absolutely different,” he says. This element of surprise is “a big part of the fun” of making generative art. 

It’s also the thing that could make it difficult to obtain copyright protection for his work in the US. Under current law, to the extent that authors are not making individualized creative choices in the production of an artwork, that work is not copyrightable. Someone like Roux—part of a resurgence of generative artists who sell their digital work as NFTs rather than as files on a thumb drive—would automatically have copyright protection for his underlying code, but probably not for the finished product. A work not protected by copyright law or another intellectual property regime is in the public domain—anyone can do just about anything they want with it. 

The purpose of copyright, per the Constitution, is to promote the progress of art. The dominant justification for intellectual property among US legal scholars today is the incentive theory, the idea that in granting authors a limited monopoly over their work, we stimulate artistic production by dangling a monetary reward. Without the promise of a payday, the theory runs, creative people would stop creating. But US copyright law does a much better job of accommodating the old model of the lonely genius toiling in solitude than it does collaboration, even collaboration between human authors. When a collaboration extends to include nonhuman authors, the law recoils. While Congress is the body that must ultimately make the decision about copyright’s outer limits, the courts addressed the question of nonhuman authorship a few years ago in the “monkey selfie” case, ruling that animals cannot hold copyright. The US Copyright Office now includes “a photograph taken by a monkey” on its list of things it refuses to register.

Similarly, the Office has long declined to register computer-generated works, like “Push-Button Bertha,” a song produced by a Datatron programmed to compose Tin Pan Alley-style tunes. It was denied copyright in 1956. The Office’s policy today is that it “will not register works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author.” That would likely exclude a good amount of the generative work being made in the crypto art space. 

And the more autonomously generated these works are, the less copyrightable they are. There’s a wide spectrum of autonomy. If the machine is merely a tool (like a camera) used by a human author, the human author won’t have a problem obtaining copyright. But scientist Stephen Thaler was recently turned down when he tried to register A Recent Entrance to Paradise, a work of visual art he says was created entirely by an AI paradigm he calls the Creativity Machine. Paradise depicts a set of train tracks disappearing into a tunnel festooned with green and purple forms that give the impression of abstracted, cascading wisteria. Parts of the picture seem as though they’ve been double exposed, with a second image swelling up within the first, giving the whole scene the effect of a kind of euphoric hallucination. 

Copyright law distinguishes between authorship and ownership, and it’s not uncommon for an entity other than the true author of a work to own its copyright. Thaler tried to register Paradise as a “work made for hire” authored by the Creativity Machine but owned by him. The Copyright Office found that the work “lacks the human authorship necessary to support a copyright claim.” Legal scholars tend to be in favor of a strictly observed binary between humans and machines, and they tend not to see any good reason to grant copyright protection to machine-generated works. (Machines, after all, don’t need incentives to create things.) 

Thaler says this reflects a Luddite perspective on what it means to be human. “It boils down to this: I’m a machine,” Thaler says, referring to himself. “I’ve invented a lot of stuff. I’ve copyrighted a lot of stuff. I’ve originated a lot of ideas in my time. And I get credit for the most part.” In Thaler’s view, the minor physical differences between the Creativity Machine and himself should not be enough to preclude the AI entity from being legally recognized as the author of its own work. 

“My feeling is that it’s a person implemented with silicon,” Thaler says. While the Creativity Machine doesn’t have every aspect—or, really, any aspect—of a human being’s anatomy, Thaler points out that anatomical perfection isn’t what makes for intellect. 

Skeptics often point out that the Creativity Machine “didn’t just invent out of thin air on its own,” Thaler says. “You had to help it,” his critics tell him. “I didn’t help it,” is Thaler’s response. “All I did was throw in some parts. Just like nature throws in some basic elements of biochemistry” to make a flesh-and-blood human being capable of creativity.

It’s unlikely that Thaler will get the acceptance of the Creativity Machine’s “humanity” that he wants from the Copyright Office. Nor should he—radically redefining our conception of what it means to be human is not a task that should fall to the Register of Copyrights, an unelected and relatively obscure government official appointed by the Librarian of Congress. But Thaler and other generative artists deserve the recognition and control that would come with at least being able to register as the authors of these works themselves. As more and more artists turn to generative code and other algorithmic tools to make their work, we should consider extending protection to the products of these methods.

To be sure, many artists in the generative art movement couldn’t care less about whether their work is eligible for copyright protection. Yet. “A lot of the people participating in the crypto space who come from a programming or coding or engineering background have that open source ethos,” says Erick Calderon, founder of the NFT platform Art Blocks. But Calderon says he sees artists start to think about protecting their images “that first time when somebody takes advantage of your work and you feel a little bit violated, where you’re just sitting there going, ‘oh, man, it would have been nice for them to have asked me.’” 

Unauthorized appropriation of an artist’s work for commercial purposes where there’s significant money at stake strikes many as unfair. And Calderon, an artist himself, sees unauthorized appropriation as both an economic and political issue. “I’d be concerned if you started a shawarma restaurant and used a Chromie Squiggle as a logo,” he says, referring to his signature generative project. “That’s not necessarily the artistic intent I had behind the Squiggles.” It’s also important to Calderon to be able to prevent his work from being used for hate speech. Without copyright, artists would have limited recourse when they saw their work being used to adorn the flag of an organization they found ideologically repugnant, or when they heard their music being used as the campaign rally soundtrack for a candidate they despised. Generative artists should be able to avail themselves of these protections too. Their work may be computer-generated, but it is not all generic—the best of it exhibits a distinct style that can be readily associated with the artist by those in the know.


There are other, less utilitarian reasons to make copyright available to generative artists. We make art for all kinds of reasons, some petty and some profound, some rational and some highly irrational. It makes sense to let artists profit from their work through copyright not because there would be no art without the cash incentive, but because money is the imperfect language the law uses to shape and communicate values. We want—or ought to want—to live in a society that values art and artists. And art that in fundamental, deeply unsettling ways challenges our understanding of what it means to be human is precisely the kind of art that our system should be endorsing, or if you prefer, incentivizing. 

There is precedent that could be useful here. We let directors—or their studios—register the films they make with the Copyright Office. Even though a film assembles the work of many different contributors—including machines and, on occasion, animals—we are comfortable assigning copyright to the “master mind” behind the film, the director who “superintended the whole work,” as one case puts it. There are hugely important differences between what film directors and generative coders do, but our model of assigning copyright to the former could provide a useful template for appropriately valuing what it is that the latter do.

Some may argue that extending copyright protection to generative art will hamper creative production overall by making it too “easy” to create a copyrightable work. A copyright troll with the right coding skills could generate a thousand images in a matter of seconds and then use them as lawsuit bait. But new technologies have always presented opportunities for trolls, and our wariness of bad actors exploiting the system should not prevent us from striving to design a copyright regime that truly lives up to its constitutional mandate.

Thaler’s perspective may seem extreme, but philosophers, environmentalists, and artists are increasingly embracing a post-human perspective to understand and navigate the crises of our time. The law, copyright law included, should help facilitate these important lines of inquiry, not stand in the way.