Whether you think they’re legitimate artistic creations or artistically derivative and plagiaristic, AI-generated images exist in a strange legal limbo that no governmental body has yet to disentangle. Yet one artist may be breaking new ground while opening up an even bigger can of worms for ownership in our progressively AI-driven world.
Kris Kashtanova, a New York-based artist and former programmer wrote to their Instagram page last week that their AI-generated art-based graphic novel titled Zarya of the Dawn received U.S. copyright registration. Noting other creators’ past failure to reach this milestone, it may be the first piece created using AI-art generators to receive such recognition from the U.S. Copyright Office.
Kashtanova wrote that their graphic novel was made using Midjourney, which is noted on the front page of the graphic novel. They said they first got the idea to register their piece as a “visual arts work” from a “friend lawyer” to try and “make a precedent.”
As noted by Ars Technica who first spotted the artist’s post, the main character of “Zarya” does bear a resemblance to Zendaya, who recently starred as Chani in Denis Villeneuve’s rendition of Dune. AI art relies on freely available images online, so basing a character off a famous actress in prompts can be a way to build consistency between images.
Gizmodo received the publicly available copyright registration for Zarya of the Dawn and found the document does not reference AI. The sole author is shown as Kashtanova, which compared to previous attempts to get copyright on AI art which named the AI as the sole author. Joel Feldman, the co-chair of the Atlanta-based law firm Greenberg Traurig’s trademark office who looked at the public filing, also noted that he was unable to find any decisions made on this property through the Copyright Office Review Board database. In an email, he did note “It is possible that something was listed in the copyright application but later deleted during examination.”
“From what I can tell, the AI issue was not squarely before the Copyright Office registration specialist and, at most, this might have been an oversight of information appearing in the deposit material but not the copyright application,” Feldman told Gizmodo.
Just last week, image hosting site Getty Images declared they wouldn’t accept anymore AI-generated images on the site, noting the ongoing questions surrounding copyright. Any AI images left on the site would be kicked off. The news followed from earlier reports that fellow image hosting site Shutterstock and Getty were quietly removing all AI images. These stock photo sites are often used by artists and photographers to monetize the use of their images online through a percentage fee on what the sites make from the content.
Kashtanova, who has written articles about selling AI-generated images, noted artists like them use sites like Shutterstock to promote their work created using AI art generators. Gizmodo reached out to Kashtanova through Instagram. Though the artist wrote they were unable to comment at this time, they supplied us with two screenshotted messages they received from Shutterstock showing that the service had removed all AI-generated content from their portfolio for a few hours—to only later restore that same content.
In response to Gizmodo’s request for comment, Shutterstock linked us to a blog post from CEO Paul Hennessy saying they were “taking steps to look at the impact AI-generated art has on our consumers and contributors.”
“There are many open questions on the copyright, licensing, rights, and ownership of synthetic content and AI-generated art,” the CEO wrote. “We need to do all that we can to not only protect the intellectual property rights of our contributors alongside the advent of this technology, but also ensure that they’re empowered to take advantage of this new creative medium.
Zarya of the Dawn is available on AI Comic Books, a site that self-proclaims to “promise a paradigm shift in storytelling.” It certainly is a paradigm shift, but one that is demanding quite a lot of heavy lifting from legal scholars and copyright holders. The U.S. Copyright Office has so far denied any application for AI to copyright the art it generates, noting before that it “lacks the human authorship necessary to support a copyright claim.”
Feldmen wrote in a recent Reuters column that one of the main issues at hand is whether the trademark office’s decision would also make AI art exempt from copyright.
But beyond thoughts of ownership there are far more pertinent questions at play. Most of the popular AI art generators use technology that compiles millions of images scraped from the internet, whether they’re photos, paintings, or even screenshots, and then use latent diffusion to compile an image based on a user’s prompt. It’s led to criticism from the art community (thanks Kotaku) that AI-image generators are plagiarizing their work, which makes the idea of monetizing the images, let alone copyrighting them, all the more problematic.
Emad Mostaque, head of Stability AI and creator of the new yet popular Stable Diffusion image generator, has shrugged off concerns these systems will hurt real artists. In an interview with BBC, Mostaque compared his tool to Microsoft Excel, which “didn’t put accountants out of work.”
That’s not to say AI-art technology should be somehow quashed. One, the artificially-composited image of a cat’s already out of the bag, and getting it back in would be next to impossible. Two, while this technology doesn’t have an actual use-case application, regular users do have a lot of fun generating wacky and oftentimes beautiful images. I personally see less harm, or at least mitigated harm, if these images are relegated to their own proprietary image hosting sites.
We’ve included a list of websites that have outright banned AI art, and a few more that should probably figure out a stated policy, and soon.