The U.S. Copyright Office has again refused to recognize an artificial intelligence as the author of a work for copyright purposes.
This renewed rejection follows Steven Thaler’s request for the Copyright Office to reconsider its earlier refusals to recognize an AI algorithm, dubbed the “Creativity Machine,” as the author of a visual work entitled A Recent Entrance to Paradise, reproduced here:
The work, which Thaler has described as the depiction of a simulated near-death experience of an artificial consciousness, was refused registration because the Copyright Office again held that the U.S. Copyright Act only recognizes copyright interest in works of “human authorship.”
In rejecting the request, the Copyright Office was unmoved by Thaler’s argument that “judicial opinions from the Gilded Age” should not control whether an algorithm can be the author of a copyrighted work. Despite being unswayed by this argument, however, the Copyright Office did not foreclose the possibility that a human involved in the creation of an AI-generated work could still be recognized as the author of the work.
There is no U.S. caselaw directly addressing the question of whether an artificial intelligence can be recognized as the author of a work for copyright purposes. The Copyright Office, however, concluded that the weight of relevant authorities supports the office’s established practice of only recognizing works created by human beings as “works of authorship,” as set out in Section 306 of the Compendium of U.S. Copyright Office Practices.
The Copyright Office’s decision to deny Thaler’s reconsideration request is primarily based on language in several century-old U.S. Supreme Court decisions such as Burrow-Giles that reference human intellectual activity when explaining the concept of copyright authorship. The Copyright Office did not respond to Thaler’s criticism that dicta from these cases, which were decided generations before the invention of the integrated circuit chip, should not be taken as authorities for refusing to recognize artificial intelligences as authors.
The Copyright Office also cited some more recent authorities, including two cases from the U.S. Court of Appeals for the Ninth Circuit rejecting non-humans as authors in other circumstances.
- Urantia Foundation v. Kristen Maaherra (1997) involved a copyright dispute between two parties who both believed that a work was divinely authored and only transcribed and compiled by mortal hands. Although the Ninth Circuit started its discussion in Urantia Foundation with the observation that “copyright laws, of course, do not expressly require ‘human’ authorship,” it ultimately held that “some element of human creativity must have occurred” for the work to be copyrightable.
- Likewise, in Naruto v. Slater (2018), a case involving a selfie allegedly taken by a crested macaque monkey named Naruto, the Ninth Circuit held that humans are the only animals with standing to sue under the Copyright Act.
The reasoning of the Ninth Circuit in Urantia Foundation and Naruto also aligns with statements about human authorship in the 1978 report produced by the National Commission on New Technological Uses of Copyrighted Works (CONTU). In that report, CONTU observed that “the eligibility of any work for protection by copyright depends not upon the device or devices used in its creation, but rather upon the presence of at least minimal human creative effort at the time the work is produced.”
Status of AI-Generated Works
How this decision will impact those seeking to protect AI-generated works under U.S. copyright law remains to be seen.
- Humans as authors of AI-generated works: Although the Copyright Office left open the possibility that a human could be recognized as the author of an AI-generated work, it did not opine on what factors would be relevant in making such a determination.
- Those wishing to be recognized as the author of an AI-generated work should be closely involved in the creation of the work (selection of training input, involvement in training process, etc.) and should document such involvement. Conversely, those wishing to challenge the validity of a copyright registration issued for an AI-generated work should seek to determine if there was sufficient human authorship in the work’s creation to justify registration.
- Infringement by AI-generated works: Whether AI-generated works are subject to copyright protection likely does not impact whether those AI-generated works are capable of infringing other works.
- In other contexts, works that are not copyrightable are still capable of infringing copyrights owned by others. One who infringes another’s copyright interest to create a derivative work forfeits any copyright interest in that derivative work; however, that new work can serve as the basis of an infringement claim despite the fact that the new work is not itself copyrightable.
- Protecting AI-generated works by contracts: To the extent AI-generated works cannot be protected under U.S. copyright law, it will become increasingly important for individuals and businesses using AI to create music, film scripts, news articles, and other works to determine to what extent contracts can be used to protect such works.
- International trends: While the U.S. Copyright Office remains skeptical of AI-authorship, the same is not necessarily true in other jurisdictions.
- For example, both the Indian Copyright Office and the Canadian Intellectual Property Office have recently recognized an AI program, RAGHAV Artificial Intelligence Painting App, as the co-author of a visual work along with its human creator Ankit Sahni.
While it is unclear if these registrations would be upheld if challenged, additional guidance on the status of AI-generated works may soon be coming from Canada as the feedback phase of the Canadian government’s Consultation on a Modern Copyright Framework for Artificial Intelligence and the Internet of Things concluded last year. What is clear is that those wishing to protect AI-generated works may need to pursue different protection strategies in different jurisdictions.
One thing is certain: The volume of new AI-generated works will increase significantly over the coming years, and, as such works become increasingly indistinguishable from human-generated works, society will need to consider not whether U.S. copyright law protects such works, but whether it should protect such works—and what would it mean to artists, creators, media and entertainment companies, content distribution platforms, and consumers if such works are ultimately covered by, or excluded from, copyright protection. Follow us on social media @PerkinsCoieLLP, and if you have any questions or comments, contact us here. Learn more about our Digital Media & Entertainment, Gaming & Sports industry group here, and check out our podcast: Innovation Unlocked: The Future of Entertainment