Court Finds AI-Generated Work Not Copyrightable for Failure to Meet "Human Authorship" Requirement—But Questions Remain

On August 18, 2023, a district court in Washington D.C., affirmed the U.S. Copyright Office's refusal to register "A Recent Entrance to Paradise," a piece of artwork created by the Creativity Machine, an AI system owned by plaintiff Stephen Thaler.

Thaler argued that the AI should be "acknowledge[d] ... as an author where it otherwise meets authorship criteria, with any copyright ownership vesting in the AI's owner[]"—that is, him. In this way, he argued that the generated artwork was akin to a "work-made-for-hire" where he owned the AI model that created the output. In rejecting Thaler's position, the court agreed with the Copyright Office that only human beings qualify as authors under U.S. copyright law, meaning that an AI-generated work like the one submitted by Thaler is not eligible for copyright protection. Thaler v. Perlmutter, No. CV 22-1564 (BAH), 2023 WL 5333236 (D.D.C. Aug. 18, 2023).

"A Recent Entrance to Paradise"

This action is the latest legal challenge by Thaler, who previously attempted to name an AI system as the sole inventor on patent applications. His effort was rejected by the United States Court of Appeals for the Federal Circuit, which found that the Patent Statute "unambiguously and directly" provides that "only a natural person can be an inventor, so AI cannot be." The Supreme Court denied certiorari of the Federal Circuit ruling; thus this remains the current state of the law on this issue—however, uncertainty remains on how much human participation will be required to meet the human inventorship requirement in the patent context. (Read our June 2023 Commentary, Generative AI-Assisted Patent Inventorship Questions Remain).

Here, Thaler argued that works created by GenAI are eligible for copyright protection, and that, as owner of the Creativity Machine, he is the owner of "A Recent Entrance to Paradise." Alternatively, he argued that he owns the work under the "work for hire" doctrine. However, the district court found that arguments concerning the copyright's hypothetical owner put "the cart before the horse," because without human involvement, no valid copyright ever existed.

In ruling against Thaler, the district court stated that "[c]opyright has never stretched so far, however, as to protect works generated by new forms of technology operating absent any guiding human hand." The district court cited the plain text and history of the Copyright Act, as well as Supreme Court and Federal Circuit precedent to support the principle that human authorship has always been a "bedrock requirement of copyright," and that a work must have a human author to be eligible for copyright protection. While Thaler owned the AI system that created the artwork, that fact alone is insufficient to meet the human authorship requirement.

As with Thaler's challenge in the patent context, this decision in this case was not surprising in view of current U.S. law and Copyright Office guidance, which among other things, noted that it is "well-established that copyright can protect only material that is the product of human creativity" and "the term 'author,' which is used in both the Constitution and the Copyright Act, excludes non-humans." Yet the Copyright Office also recognized that a work containing AI-generated material may also have sufficient human authorship to support a copyright claim, such as where a human modifies or arranges the AI-generated material in a sufficiently creative way (e.g., a collage of AI-generated images). In these cases, copyright would only protect the human-authored aspects of the work, and not the AI-generated material.

Looking forward, the devil will be in the details as courts and creatives try to determine just how much human participation will be required in order to meet the human authorship requirement. Currently, the Copyright Office relies on applicant disclosures to identify AI-generated content, explaining that copyright applicants "have a duty to disclose the inclusion of AI-generated content in a work submitted for registration and to provide a brief explanation of the human author's contributions to the work." But in many instances (particularly as technology continues to evolve), this will not as straightforward as it sounds—how much will a human being need to have contributed on the input side, both quantitatively and qualitatively, in order to claim authorship over the resulting output? In addition to open questions related to output authorship and ownership, serious questions remain as to when the output of a GenAI tool can be found to be infringing on third-party works, as numerous lawsuits have already been filed alleging that GenAI tools have "scraped" and thus unlawfully digested and copied third-party works that are used to train the tool.

Indeed, on August 30, 2023, and as part of the Copyright Office's ongoing AI Initiative, the Office published a notice of inquiry and request for comments on a variety of issues related to the intersection of copyright and AI, including "the use of copyrighted works to train AI models, the appropriate levels of transparency and disclosure with respect to the use of copyrighted works, and the legal status of AI generated outputs." The notice sets forth numerous specific AI-related questions for which the Copyright Office seeks comment, underscoring the significant challenges and uncertainty related to AI and copyright. The deadline for submitting comments to the Copyright Office is currently set for October 18, 2023.

Courts are likewise grappling with concerns related to AI-generated content, and in particular how to address such content when it shows up in legal filings. In a recent headline, lawyers were sanctioned by a court for having filed a brief written by a GenAI tool, which included false quotes and citations to non-existent court cases resulting from "hallucinations" that can sometimes occur when using GenAI. As the use of GenAI becomes increasingly prevalent, judges have begun to issue standing orders to address GenAI in written filings, including Judge Baylson (E.D. PA), Judge Subramanian (S.D.N.Y.), and Judge Vaden (Ct. Int'l Trade). The standing orders attempt to balance the potential utility of GenAI tools on the one hand, with concerns about hallucinations and protecting confidential information on the other hand.