The Copyright Office Refused to Register AI-Generated Works

The U.S. Copyright Office instituted an inquiry into a registration relating to a graphic novel that uses AI-generated artwork. Scott Hervey and Josh Escovedo  talk about this case on this episode of The Briefing by the IP Law Blog.

Watch this episode on the Weintraub YouTube channel here.

Show Notes:

Scott:
Kristina Kashtanova is the author of the graphic novel, Zarya of the Dawn. Ms. Kashtanova had previously applied for and obtained a copyright registration for the graphic novel. Sometime after the novel was registered, the Copyright office, on its own accord, instituted an inquiry into registration with a specific focus on the artwork in the novel which was generated through the use of Midjourney AI technology. After the inquiry, the Copyright office revised the registration and excluded from protection the images created by the use of the AI technology. We are going to talk about this on the next installment of the Briefing by the IP law blog

Scott:
Here is the brief history of the matter. In September, 2022 Kashtanova submitted a copyright application for her graphic novel.  Her application did not disclose that she used an AI application to create any part of the novel nor did she disclaim any portion of the work. After the application was registered, apparently the Copyright office became aware of statements made by Kashtanova in social media that she had created the graphic novel using Midjourney’s AI tool.

Josh:
Don’t you find it a bit odd that the Copyright office was watching or reviewing Kashtanova’s social media. The examiner must have had some belief that AI was somehow involved in the creation of the graphic novel.

Scott:
I never thought that the registration process at the copyright office was that in depth. I always thought it was more pro-forma, but it seems I was mistaken. In any event, based on this new information, the Copyright office determined that the original application was incorrect or substantively incorrect. The office then notified Kashtanova that it intended to cancel the registration unless she provided additional information in writing showing why the registration should not be canceled.

Josh:
As we have previously covered, the basis for the Copyright office’s refusal to register AI generate work comes from how the Copyright Act defines the scope of copyright protection.  Under the Act, a work may be registered if it qualifies as an “original work[] of authorship fixed in any tangible medium of expression. Courts interpreting the phrase “works of authorship” have uniformly limited it to the creations of human authors.

Scott:
According to the internal Copyright Offices practices, as codified in Rule 503.03, In order to be entitled to copyright registration, a work must be the product of human authorship. Works produced by mechanical processes or random selection without any contribution by a human author are not registrable.

Josh:
In responding to the Copyright office, Kashtanova’s lawyers argued that both the novel’s text and the selection and arrangement of the text and images are copyrightable since the text was written solely by Kashtanova and the text and images were arranged solely by Kashtanova. The copyright office accepted these arguments and held the text to be copyrightable as well as the arrangement of the text and images. With regard to the images, Kashtanova’s counsel argued that she used the AI platform as a creative tool, similar to how a photographer may use Adobe Photoshop.

Scott:
The decision rendered by the Copyright office goes into great detail about how Midjourney works and how Kashtanova used Midjourney. This is important to understand why the copyright office found the way it did. I am going to simplify the description a bit. Midjourney offers an artificial intelligence technology capable of generating images in response to text provided by a user. Users operate Midjourney through “prompts,” which are text commands. Prompts must start with the text “/imagine” and contain text describing what Midjourney should generate. Users also have the option to include a URL of one or more images to influence the generated output, or parameters directing Midjourney to generate an image in a particular aspect ratio or providing other functional directions.

Josh:
After a user provides Midjourney with a prompt, the technology will generate four images in response. The images are provided in a grid, and buttons underneath the grid allow users to request that Midjourney provide a higher-resolution version of an image, create new variations of an image Midjourney generate, or to generate four new images from scratch.

Scott:
Kashtanova’s counsel argued that the artist and not the machine guided the structure and content of each image. Her counsel argued “all the images in the Work were designed by Kashtanova. The visual structure of each image, the selection of the poses and points of view, and the juxtaposition of the various visual elements within each picture were consciously chosen. These creative selections are similar to a photographer’s selection of a subject, a time of day, and the angle and framing of an image. In this aspect, Kashtanova’s process in using the Midjourney tool to create the images in the work was essentially similar to the artistic process of photographers – and, as detailed below, was more intensive and creative than the effort that goes into many photographs.” And it was that “creative process” that Kashtanova applied to the creation of each image that the Copyright Office found fault with.

Josh:
First Kashtanova entered a text prompt for Midjourney. Then she would pick one or more of the outputs to further develop and would tweaked or changed the prompt as well as the other inputs provided to Midjourney” to generate new intermediate images, and ultimately the final image. Kashtanova did not claim to create any visual material herself. To obtain the final image, Kashtanova describes a process of trial-and-error, in which she provided “hundreds or thousands of descriptive prompts” to Midjourney until the “hundreds of iterations [created] as perfect a rendition of her vision as possible.

Scott:
The copyright office held that Midjourney users are not the “authors” for copyright purposes of the images the technology generates. The Supreme Court has explained, that an “author” of a copyrighted work is the one “who has actually formed the picture,” the one who acts as “the inventive or master mind.” A person who provides text prompts to Midjourney does not “actually form” the generated images and is not the “master mind” behind them. While the information in the prompt may “influence” generated image, it does not dictate a specific result. Because of the significant distance between what a user may direct Midjourney to create and the visual material Midjourney actually produces, Midjourney users lack sufficient control over generated images to be treated as the “master mind” behind them. And this lack of control over Midjourney’s specific outputs makes the Midjourney AI platform different from other tools used by artists.

Josh:
The copyright office makes an interesting comparison of the prompts to instructions given to a commissioned artist. If. Kashtanova had commissioned a visual artist to produce an image and gave that artist the same detailed instructions as she included in her prompts to the Midjourney platform, Kashtanova would not be the author of that image. Absent the legal requirements for the work to qualify as a work made for hire, the author would be the visual artist who received those instructions and determined how best to express them.

Scott:
And while Kashtanova spent a significant amount of time working with Midjourney, that effort alone does not make her the author for copyright purposes. The argument that “sweat of the brow” can be a basis for copyright protection has been rejected by courts and the Copyright Office “will not consider the amount of time, effort, or expense required to create the work” because they “have no bearing on whether a work possesses the minimum creative spark required by the Copyright Act and the Constitution,

Josh:
So, where does this leave the issue of AI and copyright.

Scott:
That’s a more nuanced question then it may seem. While the output method Kashtanova used for Midjourney does not result in copyrightable works, the office did say It is possible that other AI offerings that can generate expressive material operate differently than Midjourney does.   Based on what the Copyright Office said here, it seems that there would need to be a more of a nexis between prompts and output, once an image was generated, further prompts would need to reflect a direct control on the next image generated.  Also, the copyright office did say that had Kashtanova made substantive edits to an image generated by Midjourney, those edits could provide human authorship worth of a grant of copyright protection