The facts are not, for the most part, in dispute. Over the course of hundreds of prompts, synthetic media artist Jason Allen instructed the image synthesis generative Artificial Intelligence (AI) model, Midjourney, in what he wanted, telling it what he was imagining, like, for example, women in Victorian garb, performing in a futuristic opera house while wearing space helmets. Through prompt after prompt, Allen had Midjourney tweak the output again and again until the AI put out three final works with which Allen was satisfied. Allen then used various other tools to make further changes to the works which he then printed to canvas. He submitted all three to the Colorado State Fair fine arts competition in the "Digital Arts/Digitally Manipulated Photography" category. One of the works, "Théâtre d'Opéra Spatial" (pictured below) won first prize, the announcement of which touched off an artistic ethics debate that is one small branch of a much larger public (often heated) conversation about whether AI-generated artwork is truly "art" in its broader sense – something that embodies an ineffably human creative spark.
Allen also applied to register the work with the U.S. Copyright Office, touching off two refusals to register and a legal battle that now sits before Colorado's federal district court, to which Allen appealed the Office's refusal on September 26, 2024.1
"Théâtre d'Opéra Spatial," an AI-generated artwork, produced in response to prompts by Jason Allen in 2022.
In his appeal to the Court, Allen argues that the Copyright Office erred in refusing registration of the work without disclaimer of AI-generated content. The Copyright Office (resting on prior court rulings as well as Office rules) has consistently held that content generated by AI be disclaimed from an application as it is not the subject of original human effort and therefore not registrable. During Allen's initial application process for the work, the Copyright Examiner, who had become aware of the image's AI origin from news coverage, requested that he separate his own contribution to the work (Allen asserted that he had further edited Midjourney's work using Adobe Photoshop and other tools) and without that, refused to register it. In response to Allen's January 2023 request for reconsideration, the Copyright Office Review Board reaffirmed the prior decision, re-emphasizing that only human-authored elements of a work are eligible for copyright protection. The Office accepted that while Allen's human-made visual edits using Adobe Photoshop and the like met the threshold for original authorship, AI-generated elements created with Midjourney and Gigapixel AI (which Allen had also used) could not be copyrighted. Because Allen persisted in seeking registration of the entire work, without excluding the AI-generated portions, the Board issued a final refusal.2
In his recently filed appeal to the Colorado federal district court, Allen argued that contrary to the Copyright Office's ruling, the work had met the "widely accepted standard for originality," emphasizing that he independently visualized the image and merely used "the assistance of a machine [Midjourney]" to create his "specific image juxtaposing the Victorian age of elegance with space age wonder." The work, therefore, he said, possessed a large degree of his own creativity. Specifically, Allen said of his own efforts that:
[h]e selected the colors, the style, and the era of the artwork, and arranged the elements in the image to represent the women dressed in elegant Victorian dresses performing opera on stage, their attire presenting a juxtaposition between old-world charm and a futuristic twist. He selected and arranged the elements to depict each performer wearing a space helmet, creating a striking contrast between the classical and the sci-fi elements. Plaintiff set the stage in a grand theater, with an audience watching intently, overlooking a large circular window through which the vast expanse of the outer world is visible, adding an otherworldly ambiance to the performance.3
Allen also lamented that the Office's refusal to register copyright in his work has left him unable to profit significantly from the work or to act against infringing copies or uses of the work: "[t]he Copyright Office's refusal to register Théâtre d'Opéra Spatial has put me in a terrible position, with no recourse against others who are blatantly and repeatedly stealing my work without compensation or credit" he said in a statement. His work had been appearing all over the internet, and being made part of others' works without his permission, he stated, giving as examples an Australian artist's derivative work and one being offered by sellers on the crafting platform, Etsy.
Allen's lawsuit takes its place among other court challenges to the Copyright Office's position on AI-generated works (or portions of works). His follows AI developer Stephen Thaler's appeal of the Office's denial of copyright registration for his artwork created using AI, though in Thaler's case the work was created autonomously by Thaler's own self-programmed AI system (the "Creativity Machine"). Allen emphasized that in his own case, the output was distinguishable as not the work of an AI alone; rather, it was Allen's over 600 instructive prompts that represented the primary and extensive creative contribution to the AI-portion of the final work Théâtre d'Opéra Spatial. Allen's significant artistic control throughout the iterative creation process, he argued, meant that the entire work should be entitled to copyright protection as his own work, without disclaimer of the AI-generated part of the work.
Allen compared using Midjourney to a photographer using a camera to take pictures. "In a literal sense," Allen posited, "the camera captures and develops the photograph, yet no one would dispute the photographer's authorship."4 Similarly, Allen claimed that he "conceived of the image, developed the prompt, and made adjustments to refine the output, thereby ensuring the final image reflects his mental vision."5 The parallel to photography is interesting but one it seems likely the courts will reject as inapt. It is true that when photography first became widely available, it was condemned as not only not "art" but as spelling certain doom for visual artists – the camera was doing all the work, critics said, and the photographer contributed nothing creative other than pointing and shooting. It was years before many art institutions began recognizing photography as an art form.6 But a camera is not creating an image, rather, it is a device that concentrates variations in light and imprints upon the medium of film (or now digital equivalent) the image of the photographed subject. Such an image has for long been recognized as a creative work that is entirely a product of a human photographer's creative choices—the photographer is "drawing" an image on a fixed medium with a camera, shifting, for example, angle, light level, and placement just as a painter uses a brush in different ways. The camera is not being told "take a picture like this" and then generating images on its own.
Allen also pointed to what he noted as inconsistencies in the U.S. Copyright Office's treatment of AI-assisted works, noting that while individually hand-painted works based on AI-generated images have received registration, his own digitally created work was denied copyright simply because it was not physically recreated. Such discrepancies, Allen argued, show that the Copyright Office undervalues and misunderstands the human creativity involved in directing AI systems.
The Copyright Office has not commented on the suit; the scheduling conference is set for early next year and so no decision is imminent in the case.
Takeaways and an Afterthought
This case will, however decided, represent important further precedent as to how copyright will apply to AI-generated works or works incorporating AI-generated material. Given the bent of the cases so far, it seems likely to tutored observers that the courts will remain unwilling to order a grant of registration for a fully AI-generated work, no matter how many human prompts are behind it, and if so, the Office will continue to refuse registration for or require disclaimer of AI-generated artwork. Because of this, artists and other creative users of generative AI would be wise to ensure that they have added sufficient human creativity of the kind accepted by the Office to transform such a work into one that is fully registrable. To the extent that a work contains significant AI-generated subject matter, authors must be prepared to disclaim such content if they wish to register copyright for it.
Afterthought: Concepts of what represents "art" change over time and subsume art generated using new technologies, from the ability to create certain proprietary colors to the camera to AIs. While the generative output of an AI like Midjourney is not currently accepted as copyrightable, the written prompts to the extent that they are sufficiently original themselves might represent the "human creativity" here. So an artist hoping to copyright a work might do two things. First, they might make sure that the artwork, even if primarily AI-generated through human prompts, is something that they have contributed to directly and significantly, so that clear human authorship would permit a copyright registration (for that human part). Second, the artist could save all of the prompts used to direct the creation as their own creative work—perhaps placing them one after the other, a new kind of "found poetry." As creative written word, the prompts themselves would be registrable along with whatever of the resulting work was human-contributed. It seems possible that in time, prompts used to direct AI would become an art form of their own, and one could imagine an art gallery in which an AI-generated piece would be accompanied by selected prompts, perhaps ones the artist had found most impacted the final work. Further, the current nature of image-related generative AI means that the same set of prompts will produce quite different images in most cases. Because of this, the "poetry" of the prompts by its nature may be used to create other works, which could become part of a "portfolio" of works relating to the "prompt poetry." Given the ineffable human creative spark mentioned above, the one that drives humans to create all manner of art, and to enthusiastically employ new media and technology to do so, the Copyright Office will always be refining what of such art is by human hand and which is attributable to talented robots. We can't always be sure where they and the courts will draw the line, but we can be sure it will be fascinating to watch it unfold.
Footnotes
1. Allen v. Perlmutter, U.S. District Court for the District of Colorado, No. 1:24-cv-02665
2. Copyright Review Board Decision, https://www.copyright.gov/rulings-filings/review-board/docs/Theatre-Dopera-Spatial.pdf
3. Allen v. Perlmutter, at 5.
4. Allen v. Perlmutter, at 28.
5. Allen v. Perlmutter, at 28.
6. https://daily.jstor.org/when-photography-was-not-art/
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.