In September 2023, the U.S. Copyright Office rejected yet another attempt to secure a copyright registration for artwork which was created, in part, with generative artificial intelligence.
The Work is titled “Théâtre D’opéra Spatial”, and it was created by Jason M. Allen using a text-to-picture artificial intelligence service named Midjourney. Mr. Allen fed text prompts into the Midjourney software to set the initial scene and layout for the Work. Then, he used manual tools, such as Adobe’s Photoshop to refine the image that Midjourney created.
As shown in the September 5, 2023 denial of Allen’s second reconsideration request [HERE], the original Midjourney work and the resulting work that Allen tried to register are quite similar:
The image on the left was created by Midjourney based on Mr. Allen’s text prompts. The image on the right is after Allen edited the Midjourney image with Photoshop, adding more detail and textures.
The legal problem with registering “The Work” solely in Allen’s name is that the Midjourney software created the “base” work that Allen then modified to arrive at the final Work to be registered. This situation is not new – oftentimes a work will use a preexisting work as a base, and the person claiming copyright in the “new” work must disclaim the original work that the person did not truly “author” under the law.
As the Copyright Review Board stated:
It is the Office’s understanding that, because Midjourney does not treat text prompts as direct instructions, users may need to attempt hundreds of iterations before landing upon an image they find satisfactory. This appears to be the case for Mr. Allen, who experimented with over 600 prompts before he “select[ed] and crop[ped] out one ‘acceptable’ panel out of four potential images … (after hundreds were previously generated).” Allen Sept. Creation Explanation. As the Office described in its March guidance, “when an AI technology receives solely a prompt from a human and produces complex written, visual, or musical works in response, the ‘traditional elements of authorship’ are determined and executed by the technology—not the human user.” AI Registration Guidance, 88 Fed. Reg. at 16,192. And because the authorship in the Midjourney Image is more than de minimis, Mr. Allen must exclude it from his claim. See id. at 16,193. Because Mr. Allen has refused to limit his claim to exclude its non-human authorship elements, the Office cannot register the Work as submitted.
The full Copyright Review Board analysis is a good primer on these issues – it explains prior AI-related decisions and ties them to pre-AI case law and concepts.
Some people view this decision as too harsh, comparing it to requiring an author to disclose if spell-check features were relied upon when drafting a book. Typically, things like proofing and editing do not rise to the level of authorship, so a spell-check service is not really comparable to a text-to-picture AI engine which creates independently registrable works. However, even if the “are we going to have to disclose spell-check to the Copyright Office?” concern is remote, it is undeniable that more and more artists will rely on tools like Midjourney to start or enhance a work of art
Artists and other creators must be aware of the Copyright Office’s stance on works which were spawned using AI. You may not be eligible to claim full copyright protection for your work depending on the tools and method you used to create it. We’ve addressed similar issues in the patent field – determining whether or not generative AI is an “inventor” for purposes of registering a patent.
It has always been important to ask ownership and inventorship/creatorship questions *before* you (and your team, if applicable) begin a new work or invention. If generative AI will be part of the creative process, then it is also important to determine what rights such technology may “take” from you or otherwise prevent you from obtaining.
