Last year, Jason M. Allen won first place at the Colorado State Fair (the “Competition”) for the two-dimensional artwork entitled Théâtre D’opéra Spatial (the “Work”), which he produced with the aid of Artificial Intelligence (“AI”). Despite receiving this accolade and Allen’s arguments that he contributed significant creative elements to the AI-generated Work, his attempts to copyright the work have been unsuccessful.
Allen originally filed for copyright registration on September 21, 2022, but did not disclose that he had used an AI system to create the work. However, given the work was the first AI-generated art to win the Competition, the examiner was aware of the issue and requested more information. In response, Allen explained that he used Midjourney, which is “a text-to-picture artificial intelligence service,” but he “input numerous revisions and text prompts at least 624 times to arrive at the initial version of the image.” Then, “he used Adobe Photoshop to remove flaws and create new visual content and used Gigapixel AI to ‘upscale’ the image, increasing its resolution and size.”
As a result, the examiner asked Allen to disclaim or exclude from his copyright claim the portions of the work generated by AI. Allen did not agree to the request and continued to request copyright registration of the entire work. The Office rejected the request, stating that the work impermissibly “included ‘inextricably merged, inseparable contributions’ from both Mr. Allen and Midjourney.”
Allen requested reconsideration on January 24, 2023, “arguing that the examiner had misapplied the human authorship requirement and that public policy favored registration.” The Office again found that unless the portions Allen did not create were excluded, then copyright registration was not possible.
Allen requested a second reconsideration on July 12, 2023, based on a number of arguments. In evaluating Allen’s second request for reconsideration, the Review Board of the U.S. Copyright Office (the “Office”) began by reviewing the requirements for copyright registration. By statute, registration is available for “original works of authorship fixed in any tangible medium of expression.” Courts have determined this requires “human creation of the work,” and they routinely have rejected requests for copyright protection of “creations of non-humans.”
The Office issued guidance as to when AI-generated work is eligible for copyright registration in March 2023. According to the guidance, the Office considers “whether the ‘work’ is basically one of human authorship, with the computer [or other device] merely being an assisting instrument, or whether the traditional elements of authorship in the work (literary, artistic, or musical expression or elements of selection, arrangement, etc.) were actually conceived and executed not by man but by machine.” The Office determines “whether the AI contributions are the result of ‘mechanical reproductions’ or instead of an author’s ‘own original mental conception, to which [the author] gave visible form.’” The analysis “depend[s] on the circumstances, particularly how the AI tool operates and how it was used to create the final work.”
If a machine creates the “traditional elements of authorship,” the work will not be registered. If a work is AI-generated but also has sufficient human authorship for copyright protection, the Office will register the portion of the work created by the human.
In analyzing Allen’s Work, the Board looked at the circumstances of the work’s creation. Allen explained his process as follows:
- He “initially generated an image using Midjourney.”
- With Adobe Photoshop, he “beautify[ied] and adjust[ed] various cosmetic details/flaws/artifacts, etc.”
- Then he “upscal[ed] the image using Gigapixel AI. Allen concedes this step does not contribute to the copyrightability of the work.
The Board addressed each of Allen’s arguments for allowing copyright protection for the work.
Allen first argued that “the office ignore[d] the essential element of human creativity required to create a work using the Midjourney program,” including “enter[ing] a series of prompts, adjust[ing] the scene, select[ing] portions to focus on, and dictat[ing] the tone of the image.” Allen argued these inputs are “on par with that expressed by other types of artists and capable of Copyright protection.”
The Board determined that inputting text prompts did not make Allen the author of the Midjourney image because “the steps in that process were ultimately dependent on how the Midjourney system processed Mr. Allen’s prompts.” The Board pointed out that prompts only “influence what the system generates and are ‘interpret[ed]’ by Midjourney and ‘compared to its training data.” The prompts are not treated as direct instructions that produce a specific result. Instead, hundreds of iterations may be required before an acceptable image is produced. Therefore, the human is not creating the “traditional elements of authorship,” and any non-human-created features must be excluded from a request for copyright protection. However, Allen “refused to limit his claim to exclude its non-human authorship elements.”
Allen also argued that “the Office is placing a value judgment on the utility of various tools” and denying copyright protection for works produced by them “would result in a void of ownership.” The Board pointed out that the Office cannot exceed the laws of Congress and the limits set by the Constitution, which have been found to exclude protection for non-human works. Therefore, the fact that not all works of authorship qualify for copyright protection does “not create a ‘troubling’ void of ownership.” Further, the Copyright Act allows the Office to require disclosure of certain information, such as whether a work is AI-generated, and “[t]his requirement is not a value judgment” but rather “a recognition of the fact that ‘[h]uman authorship is a bedrock requirement of copyright.”
Next, Allen argues that “the fair use doctrine ‘would allow for registration of the work’ because it ‘allows for transformative uses of copyrighted material.” He alleges the AI-generated portion was merely “raw material” that he “transformed through his artistic contributions.” Therefore, Allen asserts the final work is eligible for copyright registration. The Board, however, points out that the “fair use” doctrine does not relate to whether a work can be copyrighted but merely whether someone has the right to use the work. Therefore, they dismissed this argument.
Lastly, Allen asserts that “requiring creators to list each tool and the proportions of the work created with the tool would have a burdensome effect if enforced uniformly.” The Board rejects this argument as well, pointing out that applicants merely need to “provide a ‘brief statement’ in the application, such as the text ‘generated by artificial intelligence,” which should not be burdensome. There is no requirement that an applicant provide a list of the tools used or how each tool was used to create the work.
In its final rejection of Allen’s request to register the work, the Board determined “that the Work contain[ed] more than a de minimus amount of content generated by  AI.” The Board found that the image from Midjourney “remains in substantial form in the final Work” and found it “is not the product of human authorship.” The Board confirmed that the AI-generated portion of the content must be disclaimed, and only the remaining content may be eligible for copyright protection. Given that Allen refused to disclaim the AI-generated portion of the work, the Board found the work ineligible for copyright protection.
This result is not surprising because it is consistent with prior results. For example, in 2018, a macaque took a selfie photo, but copyright registration was denied because monkeys are not human and thus cannot hold copyrights.
However, Allen is apparently not giving up and has indicated he will appeal the Board’s decision. We’ll see where the line is ultimately drawn as to copyright eligibility for joint creations of human and machine.