The U.S. Copyright Office’s Zarya Decision, and the Uncertain Future of AI Comics

Nathaniel Sans

New U.S. Copyright Office Initiatives on Artificial Intelligence

On March 16, the U.S. Copyright Office (USCO) announced a new artificial intelligence (AI) initiative that will include a series of public listening sessions on the use of AI in expressive works, and a notice of inquiry requesting public comment on copyright issues stemming from use of AI that USCO expects to issue this year.[1] USCO also issued a policy statement on March 16 describing its approach to the registration of works that include content generated using AI tools.[2]

USCO’s flurry of activity around the use of AI follows two significant early events in the nascent AI/copyright space:  USCO’s denial of copyright registration for the AI-generated work “A Recent Entrance to Paradise,” which is currently the subject of litigation;[3] and USCO’s decision-making on the copyright registration of the graphic novel Zarya of the Dawn (Zarya). Zarya’s author Kris Kashtanova used the AI tool Midjourney to create the art appearing in the work’s individual panels.[4] USCO’s decisions regarding those two works, combined with its pending evaluation of the AI/copyright intersection and existing social pressures related to use of AI tools to create art, create a complex environment for artists exploring the use of these tools to create comic books.

The Zarya Decision:  Background and Significance

In a February 21 letter to an attorney representing Mx. Kashtanova, the USCO said that it recognized Mx. Kashtanova as the author of “[Zarya’s] text as well as the selection, coordination, and arrangement of the Work’s written and visual elements,” but USCO declined to recognize Mx. Kashtanova as the author of the individual images (i.e., the panels) that, in aggregate, constitute Zarya.[5]

Key to the USCO’s decision was the Office’s understanding of the interaction between Mx. Kashtanova and the Midjourney AI service:  Mx. Kashtanova provided Midjourney with “prompts” that Midjourney then refined using algorithms to turn static into recognizable images.[6] USCO concluded that even though Mx. Kashtanova performed this task repeatedly on the same image and provided follow-on instructions to Midjourney to make changes to a previously-generated image, the lack of predictable output following the issuance of prompts made these prompts “ . . . closer to suggestions than orders, similar to the situation of a client who hires an artist to create an image with general directions as to its contents.”[7] Van Lindberg, the attorney who represented Kashtanova in correspondence with USCO, criticized this approach as “ . . . anthropomorphizing [Midjourney] and coming to an invalid conclusion.”[8] To Mr. Lindberg, “. . . AI tools can’t be creative. The only possible source of creativity comes from the human using the tool.”[9]

Difficult Questions Confronting Comics Creators Using AI Tools

AI tools like Midjourney or DALL-E could empower writers to bypass human comic book artists by creating their own images to accompany a written script without drawing the images themselves. Critics have attacked various elements of AI art tools, including the use of potentially copyrighted works in training datasets without compensating those works’ owners, AI-tool-users’ ability to generate images that mimic artists’ distinct styles, and the looming possibility that AI tools could cheaply perform work that otherwise would have gone to artists.[10] In the comic book space, comic book editors have publicly said they would look poorly on submissions that include AI art.[11] USCO’s split decision on the copyrightability of AI comics, coupled with the hostile reception AI art has received in the comics world, create strong disincentives for writers to experiment with AI tools when doing so carries risks of forgoing copyright protection of some aspects of their work, and social stigma in the eyes of some of their colleagues in the industry.

By cleaving the copyrightable text and compilation of images from the uncopyrightable individual images generated using Midjourney, USCO placed creators using tools like Midjourney in a difficult position. Based on the Zarya precedent, a comic creator who tells a superhero origin story in an AI-generated comic (and follows a creative process like Mx. Kashtanova’s) can receive a federal copyright protecting the story’s words and the work in toto, but USCO’s ruling on individual panels suggests that the creator cannot copyright a character’s AI-generated appearance or costume. Therefore, other creators would be free to copy the hero’s appearance to tell their own stories, which would deprive the original author of important controls over the future uses of his character. This risk of forgoing protection of a vital component of the character could serve as a potent deterrent to creators exploring the use of AI for their next comics project.

The Future is Murky

The future of AI comics is far from certain. In its March 16 policy statement, USCO stated that when it evaluates works containing AI-generated material, it will conduct a case-by-case inquiry into the operation of the AI tool “and how [the tool] was used to create the final work.”[12] The inquiry will be fact specific and apparently tailored to the specific AI technology used, so the USCO decision on Midjourney may differ from works created using different tools. Mr. Lindberg, the attorney for Mx. Kashtanova, is optimistic that the future may be brighter than it seems at present for artists using AI tools:  “I think the USCO will need to adopt a wider embrace of AI-assisted content. There is too much creative activity occurring using AI tools for that to be anything but the final conclusion.”[13]


[1] NewsNet Issue 1004, Copyright Office Launches New Artificial Intelligence Initiative, U.S. Copyright Office (Mar. 16, 2023), [] [].

[2] Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence, 88 Fed. Reg. 16190 (Mar. 16, 2023), [] [].

[3] Min Chen, A Scientist Has Filed Suit Against the U.S. Copyright Office, Arguing His A.I.-Generated Art Should Be Granted Protections, Artnet (Jan. 12, 2023), [] [].

[4] Linda Codega, An AI-Illustrated Comic Has Lost a Key Copyright Case, Gizmodo (Feb. 23, 2023), [] [].

[5] Letter from Robert J. Kasunic, Associate Register of Copyrights and Director of Registration Policy and Practice, U.S. Copyright Office, to Van Lindberg, Taylor English Duma LLP, at 1 (Feb. 21, 2023), [] [].

[6] Id. at 7-8.

[7] Id. at 8-10.

[8] Email from Van Lindberg, Taylor English Duma LLP, to author (Mar. 29, 2023) (on file with author).

[9] Id.

[10] Jo Lawson-Tancred, Will A.I. Usher In the End of Human Artists? Fear Not, Some Say, Artnet (Sept. 14, 2022), [] []; Tasha Robinson, The Wes Anderson Artbot Craze Is a Fun Trend, But It Clarifies AI Art’s Ethical Issues, Polygon (Dec. 7, 2022), [] []; Charlie Warzel, I Went Viral in the Bad Way, Atlantic (Aug. 17, 2022), [] [].

[11] Jeremy Blum, The Comics Industry Takes a Collective Stance Against AI Art Usage, CBR (Dec. 17, 2022), [] [].

[12] Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence, 88 Fed. Reg. 16190 at 16192 n.25 (Mar. 16, 2023), [] [].

[13] Email from Van Lindberg, Taylor English Duma LLP, to author (Mar. 29, 2023) (on file with author).