AI-Generated Artwork Ruled Ineligible for Copyright Due to Lack of Human Authorship
09/05/2023The evolution and rapid advancement of artificial intelligence (“AI”) is transforming many industries, but courts are just starting to examine how AI interacts with existing copyright law governing artistic works. In a groundbreaking new case handed down this summer, a federal court has now ruled that an artwork wholly generated by AI is ineligible for copyright protection.
We wrote about this case last year, when it was in its early stages (see our previous post here). In short, in 2022, scientist Stephen Thaler sought to register with the US Copyright Office (USCO) an artwork, entitled A Recent Entrance to Paradise, which had been generated by the “Creativity Machine,” an AI system created and owned by Thaler. The USCO rejected his application, citing a lack of “human involvement.” This summer, a Washington, D.C. federal judge reaffirmed the USCO’s decision, holding on summary judgment that the AI-generated work is ineligible for copyright protection.
The fundamental issue in the case was whether an artwork produced entirely by an AI system without human involvement qualifies for copyright status. In her opinion, Judge Beryl Howell emphasized that, in Thaler’s USCO application, he characterized this work as independently produced via an algorithm. Specifically, he listed the Creativity Machine as the work’s author; stated that the work was “[c]reated autonomously by machine”; and staked his own claim to the copyright based solely on his “[o]wnership of the machine,” citing the work-for-hire doctrine.
The District Court held that, in appealing the USCO’s rejection, Thaler had attempted to retroactively recharacterize the work’s creation by claiming he played a controlling role in generating the work; in doing so, he sought to add new facts to the record, including that he “provided instructions and directed his AI to create the Work,” that “the AI is entirely controlled by [him],” and that “the AI only operates at [his] direction.” But, Judge Howell noted, those statements contradicted the record that had been before the USCO, and under administrative law principles, judicial review of a final agency action like this one must be “limited to the administrative record,” and the “reviewing court should have before it neither more nor less information than did the agency when it made its decision.”
Thus, limiting her analysis to the record before the USCO, the judge opined: “this case presents only the question of whether a work generated autonomously by a computer system is eligible for copyright.” And there, she held, “Copyright has never stretched so far… as to protect works generated by new forms of technology operating absent any guiding human hand, as plaintiff urges here. Human authorship is a bedrock requirement of copyright.”
In support, the court discussed the plain text of the Copyright Act. While the word “author” is not explicitly defined in the statute, a dictionary definition indicates that the Act requires a copyrightable work to have a creator with the “capacity for intellectual, creative, or artistic labor.” The court also cited a landmark case, Burrow-Giles Lithographic Co. v. Sarony, 111 U.S. 53 (1884), which dealt with the copyrightability of a photograph; Judge Howell noted that the Supreme Court’s reasoning in Sarony was grounded in the fact that a human creator, not the camera, conceived of and designed the image, and then simply used the camera to capture it. She also reviewed other lower court precedents suggesting that works created absent human authorship (such as works created by animals, nature, or deities) should not be covered by copyright. Finally, she noted that a driving force behind the original concept of American copyright law was to incentivize human creativity; but “[n]on-human actors need no incentivization.”
Judge Powell acknowledged that the courts “are approaching new frontiers in copyright as artists put AI in their toolbox to be used in the generation of new visual and other artistic works.” She predicted “challenging questions regarding how much human input is necessary to qualify the user of an AI system as an ‘author’ of a generated work, the scope of the protection obtained over the resultant image, how to assess the originality of AI-generated works where the systems may have been trained on unknown pre-existing works, how copyright might best be used to incentivize creative works involving AI, and more.” But this case, she opined, did not need to grapple with those questions because Thaler’s USCO application had been clear that the AI system had generated this work “autonomously.”
This case illustrates the tension between our existing copyright law and the evolving art ecosystem. AI undoubtedly changes the relationship between human creativity and the actual production of a finished artwork—but this case does not answer the question of the nature and degree of human involvement required for an AI-generated work to be eligible for copyright. It only rules that the involvement cannot be zero. Indeed, this case’s outcome was significantly shaped by the way Thaler drafted his original USCO application; in the future, parties will likely strive to be more explicit about the programming and function of these AI systems, so that a court can engage in a more specific analysis of the ways in which a system’s creator and users actively shape and control the system’s output.
The case also begs the question of whose human involvement is relevant. Here, Thaler was both the programmer and the end user of the Creativity Machine. But in many situations, there are multiple humans involved in multiple stages of the creation of an AI-generated artwork; for example, a programmer might create an AI system, and then might “train” the AI system using pre-existing works by other artists, and then a separate person might be the one who uses the system to create a final product. How is our copyright regime to account for the creative input of all of these individuals, who may not even know each other? Indeed, litigation is already underway in several suits filed by artists and writers who claim their work has been used to train AI systems (see here, here, and here for examples), while another artist is also appealing the USCO’s denial of his copyright application (on grounds similar to those cited in Thaler’s) for an artwork he generated using an AI software called Midjourney. It seems likely that at some point, copyright law may need to be legislatively updated in order to address the problems posed by the unique nature of AI technology; but until Congress acts, the federal courts will have the difficult task of trying to apply age-old copyright law principles to a new and complex technology.
Press reports indicate that Thaler intends to appeal this decision, so we’ll continue to follow the case.
Note: This blog post was co-authored by Jacqueline Jakimowicz and Kate Lucas. Jacquie was a legal intern at Grossman LLP this summer and is now in her 3L year at the University of Miami Law School.
Art Law Blog