Is Artificial Intelligence Copyrightable? A Report on the Copyright Office’s AI Initiative

August 26, 2024

By Ana Larsen

As generative Artificial Intelligence (AI) has increasingly integrated into daily life, its role has evolved from being an assistive tool, such as a translation aid, into a service capable of rendering its own creative work. Although it is reliant on using human-made work to train and can be prone to error, generative AI has already proven to be an incredibly efficient and interesting tool across many fields. One example of its successful integration in the art world is the 2016 collaborative artistic project “The Next Rembrandt.”

A collaboration between art historians, data scientists, developers, engineers, and generative AI, The Next Rembrandt project analyzed Rembrandt’s existing works in order to render an entirely new painting in his style.

However, in cases like “The Next Rembrandt” to what extent is the work copyrightable? Can art made by AI be copyrighted, and if not, what extent of human input is required to copyright a piece of art?

Due to the ambiguity surrounding what constitutes authorship, these issues have been brought to U.S. courts and agencies several times, most significantly in Thaler v. Perlmutter.

Landmark Copyright case: Thaler v Perlmutter

(For an in-depth case review of Thaler v. Perlmutter, see CFAL’s previous article).

Stephen Thaler first applied for a copyright claim with the Copyright Office on November 3, 2018. The application listed the “Creativity Machine” as the author of the work and listed Thaler as claimant with a transfer statement of “ownership of the machine.” ^{His claim was founded on the basis of registering the “computer generated work as a work-for-hire to the owner of the Creativity Machine.” On August 12, 2019, the Copyright Office rejected his application claiming that the work “lacks the human authorship necessary to support a copyright claim.”}

On September 23, 2019, Thaler requested that the Copyright Office reconsider his application for copyright ownership, claiming that the requirement of human authorship is unconstitutional. After a re-evaluation, the Copyright Office again concluded that Thaler “provided no evidence on sufficient creative input or intervention by a human author in the Work” and therefore the Work did not have the necessary component of human authorship to sustain a copyright claim.

Again, on May 27, 2020 Thaler requested a second reconsideration. Bearing the same arguments as his first request, Thaler furthered his application by emphasizing public policy, and claiming that by copyrighting machine-generated works, the “underlying goals of copyright law” would be furthered. ^{The Copyright Office refused the reconsideration again on February 14, 2022 citing its Compendium which states that “the Office will refuse to register a claim if it determines that a human being did not create the work.”}

As a result of the repeated rejection, Thaler filed a complaint on June 2, 2022, against the Register of the Copyright Office, Shira Perlmutter. ^{A key point in Thaler’s complaint was that the Copyright Office’s denial of his registration contradicted “the plain language of the Copyright Act (“Act”), … the statutory purpose of the Act, … [and] the Constitutional mandate to promote the progress of science.”}

On August 18, 2023 Judge Howell ordered that Thaler’s motion be denied, concluding that the Office acted appropriately in its denial “for a work created absent of any human involvement”.^{Thaler filed an appeal against Judge Howell’s decision on October 18, 2023. Both sides have filed briefs in the meantime. The case, which is still currently open, has a scheduled oral argument for September 19, 2024.}

As is evident in Thaler v Perlmutter, the Copyright Office is a crucial player in copyright law, but when novel concepts like Artificial Intelligence are introduced to the art world, how adaptive can it be?

What is the Copyright Office

The U.S. Copyright Office has been part of the Library of Congress since 1870^{and its role includes: examining and registering copyright claims, recording information about copyright ownership, and assisting Congress in developing regulations concerning copyright law. The Copyright Office also works closely with the public, both through educational outreach as well as relying on public feedback. Its relationship with the public is crucial in developing copyright law regulations, as public input is a key component in how these regulations are determined.}

On March 16, 2023, as part of its duty to investigate public concerns on updating copyright regulations, the Copyright Office launched its Artificial Intelligence Initiative.

A Timeline of the Copyright Office’s AI Initiative:

• March 16, 2023: The Copyright Office launched its Artificial Intelligence Initiative through its NewsNet Issue 1004. According to the Office, the launch was “in direct response to the recent striking advances in generative AI technologies and their rapidly growing use by individuals and businesses” as well as public and congressional requests to examine these issues with regards to copyright. ^{Along with launching the initiative, the Copyright Office also issued a new registration guidance which specified applicants’ responsibility to disclose any role generative AI has in works submitted for copyright registration. The registration guidance further emphasizes the human authorship requirement in its explanation of what balance of human input and AI-generative material can be copyrightable.}

• April-May 2023: Throughout the spring of 2023, the Copyright Office held four public virtual listening sessions. As described in Office’s report, the purpose of these sessions was to listen to the public’s concerns, expectations, and questions regarding copyright regulations and generative AI. ^{The first listening session on literary works set the precedent of questions that were to be addressed throughout the sessions. These guiding questions included: “How does current law apply? Should it be changed? [H]ow will the copyright community, from creators to users, be impacted?” The remaining three sessions addressed these concerns through the lenses of: Visual Arts, Audiovisual Works, and Music and Sound Recordings. On the 18th of July 2023, the Copyright Office published an informal report summarizing the listening sessions. The report revealed three common threads throughout the sessions: 1) disagreement about whether (or how) training generative AI on existing copyrighted works could be considered fair use, 2) an interest in improving public understanding on how generative AI produces works, specifically tracking relationships between input and output of AI generative machines, and 3) clarity on the registration guidance published by the Office on 16 March 2023. Although the July report served as an informal precedent to the later published Notice of Inquiry, it is a telling sign that a large public (over 4,100 participants) is concerned with how generative AI should be regulated.}

• June 28, 2023: Registration Guidance for Works Containing AI- Generated Content.  After the listening sessions determined that the public wanted clarity on copyright registration of works containing AI generated material, the Copyright Office held its first Online Webinar to review the Registration Guidance published March 16, 2023. Up until that point the Copyright Office had received less than 100 claims of works which incorporated generative AI, but expected this number to substantially increase. ^{With this context, the webinar reviewed the Registration Guidance, in which two main points were emphasized: 1) the duty for artists to disclose any “appreciable amount” of AI generated work, and 2) that this can be done through a simple statement, similar to which one would disclose other unclaimable material. Along with these reiterations, the Webinar included several examples to further examine the intricacies of the guidance policies.}

• July 26, 2023: International Copyright Issues and Artificial Intelligence In conjunction with the June 28 webinar, the Office held a virtual presentation on global issues surrounding generative AI. Four panelists discussed the global impact of generative AI and copyright regulation through presentations on: AI developments across Asia, the proposed Artificial Intelligence Act in the EU, data mining in the global south, and AI issues related to language bias.^{. Overall, the summer webinar series of the Copyright Office’s AI initiative provided a wealth of information from regional experts, as well as an opportunity for the public to voice concerns and questions regarding policy issues and AI.}

• August 30, 2023: Notice of Inquiry The Copyright Office published its Notice of Inquiry on Generative AI in the Federal Register to elicit public opinion on a multitude of issues. The Notice’s inquiries to the public included comments on: “(1) the use of copyrighted works to train AI models: (2) the copyrightability of material generated using AI systems; (3) potential liability for infringing works generated using AI systems; and (4) the treatment of generative AI outputs that imitate the identity or style of human artists”. ^{Over 10,000 comments were submitted by the December 6 deadline. Comments came from individuals, companies, and federal organizations, and can be found on the Regulations.gov website.}

• December 2023- February 2024: Beginning the winter of 2023, Shira Perlmutter, the Register of Copyrights and Director of the Copyright Office provided several updates to Congress on the status of the AI initiative including a Letter to USPTO & USCO on National Commission on AI (copyright.gov) and Update to Congress. Given that the final goal of the Copyright Office’s AI initiative is to issue reports to Congress with suggestions for improved federal regulation, regularly providing Congress with updates preemptively sets the stage for the final reports.

• July 31, 2024: Copyright and Artificial Intelligence, Part 1 Digital Replicas Report As a result of the AI initiative and the large amount of public opinion, the Copyright Office published its first report on Copyright and AI. Taking public input into consideration, the Office concluded in its suggestions to Congress that although current federal statutes on artistic style and copyright are adequate, urgent updates to federal reform are necessary in regards to other issues created by digital replicas.

Current Federal Statutes:

In its report, the Office acknowledged the areas in which current legislation serves adequately. Despite some public concern over protecting artistic style, the Office determined that although the Copyright Act does not directly address artistic style as a separate part of the work, the act could be applied in instances in which the output happens to replicate other protectible portions of the work, as well as the artistic style. ^{The Lanham Act also provides certain solutions in this respect.}

Current legislation with regard to Copyright regulation on AI lacks in several other respects however. Many states have right of privacy and right of publicity laws, however the limits in who they protect and to which degree are not sufficient enough to adapt to the rapid developments of generative AI. No federal statute currently exists with the sole purpose to protect the use of an individual’s likeness, image, or voice, and the existing protections provided under the Copyright Act, Lanham Act, Communications Act, Federal Trade Commission Act are too narrow to account for AI innovation.

Risks associated with Current Legislation:

As stated above, aside from protecting artistic style, current legislation is not sufficient to protect the public from other issues that can arise out of digital replication. For example, within the creative sector, unauthorized cloning of imagery, voice, and likeness, (even as far as using digital replica extras instead of background actors), can threaten individuals’ income and livelihoods. Not only are creatives at risk, but generative AI’s ability to create digital replicas of anyone and anything leaves the public vulnerable to misinformation, misrepresentation, and calumny. As a result, urgent federal regulation is required to protect the public.

Proposed Federal Statutes:

Taking into account public input, the Copyright Office report includes a proposed New Digital Replica Right to Congress. The essential components of this new act are: “(1) the definition of “digital replica;” (2) the persons protected; (3) the term of protection; (4) prohibited acts; (5) secondary liability; (6) licenses and assignments; (7) accommodation of First Amendment concerns; (8) remedies; and (9) interaction with state laws.” ^{In short, with these key elements in mind, the Office recommends to Congress to establish a federal right that serves all individuals throughout their life by protecting them from the distribution of unauthorized digital replicas.}

Conclusion

Within the context of cases such as Thaler v. Perlmutter, it is clear that copyright policy with regards to generative AI needs to be clarified, regulated, and easily adaptable. The Copyright Office’s AI initiative successfully drew a mass amount of public input, as well as created an informative discourse with experts on subjects within AI policy. Although as of today, only Part 1 of its reports has been published, the Copyright Office successfully distilled a mass amount of public concerns into precise recommendations for Congress. Given the rapid developments in AI, it is now crucial that Congress seriously considers the Copyright Office’s reports and implements updated Federal regulations.

Suggested Readings:

• Ivan Moreno, Art Requires Human Input, DC Circ. Told in AI Copyright Row, Law360 (March 7, 2024) available at https://www.law360.com/articles/1810961/art-requires-human-input-dc-circ-told-in-ai-copyright-row
• Blake Brittain, Computer Scientist makes case for AI-generated copyrights in US appeal, Reuters (January 23, 2024) https://www.reuters.com/legal/litigation/computer-scientist-makes-case-ai-generated-copyrights-us-appeal-2024-01-23/
• Steve Schlackman, Who holds the Copyright in AI Created Art, Art Journal (September 29, 2020) available at https://journal.atp.art/the-next-rembrandt-who-holds-the-copyright-in-computer-generated-art/

About the Author:

Ana Larsen currently works as an International Arbitration paralegal in Washington D.C. She graduated in 2023 with a BA in Art History from Washington University in Saint Louis, and a second major in Design and a minor in Spanish.

Bibliography:

Facebook
Twitter
Pinterest
LinkedIn
Tumblr
Email