Intellectual property is theft
Intellectual property law is failing to protect artist livelihoods and art as a public good.
I'd thought maybe some of the global discussions around intellectual property had cooled down a little, what with various forms of commercial streaming displacing a lot of digital media piracy. But now recent flashpoints around AI-assisted art enabled by the proliferation of tools like Midjourney, Stable Diffusion (visual art), and ChatGPT (writing) are raising new questions about intellectual property.
These tools are typically trained through a complicated analysis of vast quantities of art published by people on the internet. When someone types a prompt into one of these tools and generates an intricate digital painting of fantasy cities, alien landscapes, and impossible creatures, is it art? If it is art, who is the artist? The person who fine-tuned a written prompt that would produce this picture? The AI tool itself? Or the untold legions of artists whose work was used as training data? And what of those artists behind the training data? Assuming their work was collected by web crawlers without their consent, was their work stolen? And, if so, by whom—the end-user of that AI tool, the AI tool itself, or the data scientists who added that art to the training data in the first place?
Yes, it's art
For me, “Is this art?” is the easy part, and independent of the other questions. The answer is yes. I don't have a philosophical proof for this, but I do have a few guiding principles and examples to make my case. Firstly, the act of crafting the right prompt for an AI tool like this is a creative act. It involves experimentation, spontaneity, and communicating the ineffible through a set of visual and stylistic descriptors. We know it involves expertise, because it can be done inexpertly; someone new to writing these prompts might often generate an image that contains all the things they described, but not at all in the style or tone they intended, or marred by unwanted, districting details, like a human hand with an improbable number of fingers.
While art created in this way is art, it is also clearly derivative of prior artwork. It's hard to escape evidence of this. Hauntingly, the graphical output of these tools sometimes features indecipherable signatures imitating artist signatures in the training data, not unlike Frankenstein's monster newly assembled from a hodgepodge of corpses and struggling to speak his name for the first time.
Art derived from the artwork of others is still art. This is not to say that it is always ethical, or that it is always “good art” by any particular measure.
One of my favorite examples of wholly derivative art is The Grey Album by Danger Mouse, which consists of the acapellas from The Black Album by rapper Jay-Z accompanied by backing tracks built entirely from samples of The Beatles' self-titled 1968 album popularly known as the “White Album.” Every sound on The Grey Album comes from these two other albums, and at the same time it is unambiguously an original artistic work. If you've heard The Grey Album that doesn't really mean you've heard The Black Album, and definitely doesn't mean you've heard The Beatles. The beats are compelling, and they highlight different aspects of Jay-Z's vocals than you might notice with a different backing track. Not just anyone could make what Danger Mouse did given access to the samples he used. It's art. It's also, legally speaking, an unauthorized use of the recordings it samples and almost certainly a violation of intellectual property law. The Grey Album began as a personal project not intended for a wide, commercial release, but gained international attention online. Ultimately, Jay-Z and the surviving Beatles approved of The Grey Album, but EMI, the corporation that controlled the cought of Beatles recordings, threatened legal action. There is some argument to be made that the use of unauthorized samples in original works could be covered by the “Fair Use” doctrine that outlines some limited exceptions in US copyright law, but this is very much a legal gray area shaped by evolving case law. I suspect The Grey Album wouldn't have great chances in court, but since the album surfaced in 2004 it has never landed there. So:
✅ It is art.
❌ It is probably not legal.
❓ Is it ethical? I don't know. As far as I can tell it didn't hurt anyone.
I don't think Jay-Z or The Beatles can be said to have suffered financially from people downloading The Grey Album. But what of the more general case? What damage is caused by the unauthorized consumption of art?
You wouldn't steal a car
A 2004 public service announcement produced by industry associations infamously compared movie piracy by home viewers to automobile theft, stating unambiguously that “Downloading pirated movies is stealing.” The PSA lives on in public memory as an object of mockery because of an obvious flaw in its central comparisons. Most people tend to think that what makes it “wrong” to steal a car isn't that you have a car you didn't have before, but that because of you, someone else doesn't have that car anymore. When you download a pirated movie, no one else is deprived of that movie. Everyone who had that movie before still has it now. But now you have it too.
This is one of the fundamental issues of intellectual property as a concept; it relies too much upon the principles of physical property, but it plainly doesn't play by the same natural laws as anything you could physically hold; you can give it to someone without it leaving your possession and it is constantly multiplying and moving all around us. If someone takes it from you, you still have it. And once it has multiplied enough, it is effectively indestructible.
When these industry associations present a more developed argument against copyright infringement, they sometimes argue that media piracy is “not a victimless crime” because it does deprive people of revenue. This point is debatable and can be subjected to empirical scrutiny.
Anecdotally, we know that at least some people who download a pirated movie simply wouldn't watch it if they couldn't pirate it—because they're broke and can't afford it, because it wasn't officially released in their region or language, or because it's rare, out-of-print, and not on streaming services.
We also know of times when media piracy has driven discovery and sales. The musician Benn Jordan, for example, has said that in the pre-Spotify days of iTunes, frustrated in his attempts to secure a cut of the profits from his music being sold on the new digital music platform, he uploaded his own music to torrent trackers, with a note to listeners containing a URL through which they could pay him directly. This created a new audience and a new revenue stream for the musician—more revenue, in fact, than the musician was getting from iTunes sales at the time. Perhaps ironically, this model of audience engagement is less viable today because so many music listeners are already buying subscriptions to music streaming services that pass unlivable and ever-shrinking shares of their profits to musicians.
So we know that, in individual cases, acts of media piracy have been neutral or even positive in their impact on artist revenues. Certainly there must be times when someone chooses not to but media they could afford because they can download pirated copies instead. So the question that can be investigated empirically is whether the net effect on artist revenues is positive, negative, or negligible. I haven't done that work or deeply read existing research in this area, so I won't assume an outcome here.
When you “steal” a movie by downloading a pirated copy, no one loses the movie. Whether artists lose revenue is somewhat unclear. But what is definitely lost is distributor control, the ability of the company that sells and licenses that movie to control the circumstances, to monetize the platforms that mediate authorized access to that movie and to target you for advertising through those platforms, to passively collect demographic data about who is watching the movie and use it to adjust their business model.
So who suffers from this crime?
✅ The ability of distributors to exercise control over your media consumption suffers.
❌ Benn Jordan apparently didn't suffer, but instead materially benefited from the piracy of his own musical recordings.
❓ Do artists in general suffer? It's something we can investigate through data.
The mythical artist
But hang on… A lot of the examples I've given regarding media piracy involve Hollywood movies. Who exactly is the artist deserving of payment when a Hollywood movie is made? A lot of film critics assume the primary creative force behind a movie is typically the director. But some academic citation formats would have you credit the executive producer first. Certainly the screenwriter plays some part. And the actors too can influence the tone and texture of a movie and sometimes even make changes to the plot. In fact a typical Hollywood movie is the creative product of a whole army of people who work at the studio; that's why the end credits are so long. And who really can say that any of the people named in that list didn't contribute artistically to the movie you saw? Would that climactic scene have felt exactly the same without the work of a competent hair stylist? Would the actors have delivered exactly the same performance if the set weren't adequately catered?
When you watch a movie in a movie theater, you don't directly pay each of those artists listed in the end credits. For artistic work under some circumstances we have established indirect means of funding. In this case you pay the movie theater, the movie theater uses ticket sales to pay some distributor, that distributor bought the distribution rights to the movie (a subset of the intellectual property rights for that movie) from the movie studio, and that studio hopefully paid all the people who worked on it, either through a percentage of revenues (for producers and top-billing actors, maybe), a lump sum negotiated before production, an ongoing contract, or meager hourly wages. Though if there were interns on set, they may have been “paid” in experience alone.
In fact, most media consumption today relies on some kind of indirect scheme to pay the artist. When your access to an artwork is mediated by institutions, those institutions determine what portion of subscription fees, ticket sales, ad revenues, merchandise profits, etc. are paid to the artist. What that portion is and how it is determined is rarely transparent to the audience or even to the artist. And anecdotally, this portion seems to be shrinking all the time in most cases.
A few rules to live by
- The world is better with art in it. That is to say, art is a public good.
- Everyone deserves a basic standard of living that includes (among other things) adequate food, shelter, and healthcare, and access to art.
- Artists have more time to make art when they don't have to do a lot of other things to maintain a basic standard of living.
The above are axioms I hold to be obvious. I think a lot of people in the world probably agree with them. Taken together, they also demand a radical rejection of intellectual property as it is currently conceived.
We live in a system where (unless you're born rich) a basic standard of living has to be “earned” through work that has been assigned a material value. In theory, the work of creating art can be assigned that material value. But it doesn't happen automatically. If you just decide to paint something, or take a carefully composed photograph, or play music, or write something, outside of any kind of contract, you will not be paid just by virtue of having made your art. You might get paid if you can find a way to sell your art, and if you do there's no guarantee that you will bring in enough money to live on or even to pay for the materials you used to make your art. To make art in this world is a good in itself, but it is also a gamble with the time and money we use to maintain a basic standard of living.
I often hear that “artists deserve to be paid.” I agree but didn't include it in my axioms above because I think that artists deserve to be paid because everyone deserves a basic standard of living and when that has to be bought with money gained through so many hours of work per week, any artist who isn't already wealthy is risking their standard of living by making their work. They could be using their time and their material resources to do work that pays them without that extra step of selling what they've already made and hoping someone buys it—whether that's the audience directly buying from the artist, or some institution agreeing to exhibit the artist's work and mediate the process of selling it.
What's so threatening about the idea that AI-assisted art tools are reassembling uncredited artwork into new, monetizable images is that in obscuring the artists behind the training data, the creators and users of these tools have cut off any chance of those artists being indirectly paid for the sale of this new art they unwillingly helped to create. It's a case that should have been addressed, I think, through consent and attribution mechanisms. But it is also a case that existing intellectual property law is unequipped to address. Existing law has not even really settled the legitimacy of derivative artworks in general, or when such works do or don't require someone's permission. This does not even touch on works that draw on thousands or even millions of other artists' work through an automated tool.
The new patronage
One very old model for bringing art into the world while providing artists with a basic standard of living is the patronage system. The storied patrons medieval Venice etc. were exceptionally wealthy people who contracted artists to make big, impressive works that were both enjoyable for the patron and a means of projecting status in high society.
Today we have things like Patreon or Substack where a bunch of people who might not be extraordinarily wealthy make small, recurring donations to an artist whose work they want to encourage, and, perhaps with the addition of some one-time sales through platforms like Bandcamp or commissions sold via direct communication and online payments systems, those donations will add up to enough for the artist to live on. Like patrons of old, Patreon donors might receive some exclusive benefits to encourage their financial support, like access to some work that is not publicly available. The exclusivity of that access often implicitly relies on intellectual property law.
I respect artists who use platforms Patreon and Bandcamp to financially enable their work. And I respect people who may not be especially wealthy but choose to support the art they enjoy through those platforms. It's sometimes pitched as a way of democratizing the distribution of art and cutting out opaque institutions with conflicting interests. I think that's an admirable thing. But it's not The Solution to our problem. To get paid this way, artists still have to do work to grow a following on Patreon or Bandcamp, when they could be working on their art itself. And it's skilled work, work most artists (most people, really) are not just naturally good at doing. It also tends to imply walling off at least a little of the art exclusively for people who can and do pay a little money for it, to entice them to keep paying. I consider that justifiable. But I'd rather we didn't have to do it that way.
Reclaiming the Creative Commons
The Creative Commons is a legal project that crafts “some rights reserved” copyright licenses written in the appropriate legalese, which people can apply to their work to explicitly allow the public to do some things with it that are implicitly disallowed by traditional “all rights reserved” copyright licenses. These licenses are meant to enable the creation of a new “commons,” a new pool of creatlive work that the public had a right to use, enjoy, share, and maybe reinterpret with free restrictions.
Creative Commons licenses somewhat paradoxically rely on existing copyright law to provide legal enforceability for their “copyleft” measures. If a litigious publisher were to allow an author to release their work under a Creative Commons license, but later sue another party for activities explicitly allowed by that license (say, non-commercial distribution of copies of that work retaining full attribution, which the Creative Commons licenses generally allow), a legal defense would rely on the author's copyright claim to give authority to the terms of the Creative Commons license they chose. It is the author's exclusive copyright claim that authorizes them to give away some of those rights implied by copyright to the public. In this sense the Creative Commons project strikes a compromise between building toward the free culture we could have and acknowledging the legality of the intellectual feudalism we do have.
But maybe the greatest weakness of the Creative Commons project is that in practice it places the onus on writers, musicians, filmmakers, artists, and “creators” of all stripes, as original copyright holders, to invest their own work in this enterprise. To trust that ceding to the public some of these rights granted to the author by intellectual property law will not threaten their livelihoods. To take a leap of faith. But in this economic landscape, will anyone be there to catch them?
The scope of the Creative Commons project is to offer alternatives that could help mitigate the public access to art part of our conundrum, but it doesn't directly address the part about guaranteeing artists access to a basic standard of living.
What is it good for?
So now I have to ask… If the current conception of intellectual property is not equipped to enable crediting and paying artists whose work contributes to the training of AI art tools used in making monetized artwork, how is it helping? If intellectual property law cannot guarantee artists a basic standard of living, and only protects their access to one inasmuch as the artist can figure out how to leverage it to sell their work or negotiate a pittance from some mediating institution, what end is it achieving? If the alternative to this system of intellectual property is a system where the livelihood of the artist still depends on monetizing the art, but the artist has no right whatsoever to dictate the terms of that monetization, then intellectual property law is a thin, arbitrary, and highly conditional protection for artists that better serves mediating institutions. But if the alternative is a system where the artist enjoys a decent standard of living independent of any monetization of the artwork, then intellectual property becomes a kind of theft, a siloing off of what could be our creative commons of artworks in public dialog with each other into paywalled gardens and the hands of gatekeepers.
Paths forward
So how do we build that alternative, the world in which the “necessary evil” of intellectual property law becomes more plainly unnecessary? The most obvious answer is a standard of living unconditionally guaranteed to all.
But I think there are steps we as a society can take to a world beyond intellectual property even within the capitalist economies we inhabit. For one thing, we could dramatically cut the out-of-control longevity of copyright terms. Under present rules, for example, the 1927 science-fiction movie Metropolis only entered the public domain in the U.S. in 2023, and will not enter the public domain in the European Union until 2047. Art in the public domain, like the works of Mary Shelley or El Greco, can be endlessly republished, reinterpreted, remixed, parodied, and pastiched by anyone and without authorization. This body of classic art that belongs to all in common is fertile ground for new and innovative derivative works. Keeping the copyright of artworks tied up in literary estates 70 years after the author has died, and potentially well over 100 years since initial publication, stunts the growth of the public domain, to the point that there is very little public domain art most of us would recognize as modern. We could act to expand that public domain in short order with a change to copyright terms.
We could start forcing publishers, record labels, streaming services, movie distributors, etc. to pay artists some minimum portion of revenues for their work. We could begin to impose some Creative Commons-like terms (like allowing non-commercial redistribution, perhaps after a very short exclusivity period at initial release) upon those same companies, and mandate that they provide lendable copies of digitally published works of any medium to public institutions like libraries for a reasonable fee.
But we should really do more than chip away at the edges of copyright law. We could and should also collectively invest in public art as a public good. At one time, the U.S. government attempted to mitigate the impact of the Great Depression upon arts, infrastructure, and the culture of democracy through initiatives like the Works Progress Administration in which they directly hired unemployed people to build public infrastructure and create public art of various kinds. It was by no means a perfect system but it points the way to what we should be doing: collectively investing in artists so they can create art without having to hustle for a living.
To that end, any collective investment that reduces the pressure on everyone to “earn a living” will of course support artists as well. Public housing and single-payer healthcare are examples. A properly implemented universal basic income (UBI) that does not displace existing social services also has the potential to vastly increase public creativity.
A better world, a world with more art in it, is possible, and there are steps we could take now to make it a reality.
The open conversation
I've hesitated to share this post because I know I'm just scratching the surface of this problem. How can we best support both artist livelihoods and art as a freely shared public good? My hope is that even talking about it in a relatively surface-level way like this enough, and in enough places, will encourage a productive level of public discourse on the subject. Because something has to change.
Where not otherwise noted, the content of this blog is written by Dominique Cyprès and licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.