Signs of Life: The Current State of Generative Content and Copyright Protection

With the recent rise in Artificial Intelligence (“AI”) and its broadening use by the public, questions have arisen regarding the applicability of copyright law over both the content it sources and creates. While the answers remain unclear as this technology rapidly updates, there have been recent legal developments which will shape how we deal with this content.

AI Use of Copyrighted Material

“Generative” programs like Open AI’s ChatGPT and DALL-E “generate new images, texts, and other content (or “outputs”) in response to a user’s textual prompts (or “inputs”).”¹ The programs, however, do not rely solely on the information provided by the user. Open AI’s programs are trained on internet sources––previously limited to information prior to September 2021, but as of September 27, 2023, ChatGPT can now access the internet freely with no time limit on its data.² Even before this expansion, there had been controversy over Open AI’s sourcing of material.³ There have been multiple lawsuits initiated by authors against OpenAI including comedian Sarah Silverman who alleges that ChatGPT was trained using an unauthorized or pirated copy of her autobiography.⁴ These cases could have a significant impact on fair use doctrine and those outcomes will certainly have an effect on the future of AI content and machine learning.⁵ While the results of these cases are yet to be determined, courts and the United States Copyright Office have begun issuing rulings and guidance about the availability of copyright protection for AI generated works.

Registrability of AI Generated Works

Historical Background

The Copyright Office has been wrestling with the issue of computer authorship since at least 1965, when the Office observed that the determining question is “whether the ‘work’ is basically one of human authorship, with the computer merely being an assisting instrument . . . .”⁶ While this sentiment was used to prevent the categorical denial of copyright over content created with computers, the issue of AI was still incredibly new and any discussion of the matter was speculative.⁷

Recent Developments

Fast-forward over fifty years, and the discussion around AI has become much more concrete. The Copyright Office has begun making determinations on applications for copyrighting AI generated content. In 2022, the Office publicly addressed the issue in two cases. The first involved a piece of artwork which had been “autonomously created by a computer algorithm,” and there the Copyright Office denied registration because it lacked “any creative input or intervention from a human author.”⁸ This requirement of human authorship, the Office determined, was well established in statute, judicial precedent, and Copyright Office practice.⁹

It is unclear, however, what the threshold of human involvement is for a work to be considered registerable. This issue can be highlighted by the other 2022 application where the Office was examining a graphic novel which had words written by a human and art created with Midjourney, another AI system. There, the Office found that the “human-authored text and human selection and arrangement of the text and images” was protectable, but the AI images themselves were not because the “human author lacks sufficient creative control over the AI-generated components of [the] work,” and is therefore “not the ‘author’ of those components for copyright purpose.”¹⁰

To further clarify its position, the Copyright Office released guidance for the registration of works containing AI generated content in March of 2023.¹¹ There, it elaborated on the longstanding requirement of human authorship and explained how that requirement would be applied.¹² In drawing a line between human and machine authorship, the Office stated that “[w]hen an AI technology determines the expressive elements of its output, the generated material is not the product of human authorship.”¹³ Some may think then that by providing more specific expressive information in their prompts, users could satisfy the requirement, but the Office does not appear receptive to that concept, noting that while “iterative ‘feedback’” from users may provide a “greater [human] influence over the output, the AI technology is what determines how to implement those additional instructions.”¹⁴ It therefore stands to reason that no level of directional control over the AI content would satisfy the Copyright Office’s requirement of human authorship. So, what should artists and creators do? One solution would be to simply create the works themselves without AI tools, thereby eliminating the question altogether. Another would be to utilize the AI content in a way that exerts human control over the expressive elements, much like in the way the “selection and arrangement of the text and images” in the aforementioned graphic novel case was found to be protectable.¹⁵ The Copyright Office remains adamant, though, that the availability of protection is limited only to human expressions utilizing the generated content and such protection exists “independent of . . . the copyright status of the AI-generated material itself.”¹⁶

As it stands, the role of AI generated content within copyright law is still in its early stages and is likely to shift as the technology continues to evolve. At present, artists who are concerned about the protectability of their works might be better off avoiding the technology altogether, or, at the very least, be wary that the protection of their work will not encompass the generated content in and of itself. Additionally, the pending cases involving OpenAI’s sourcing of information along with the program’s now unlimited access to the internet could spell trouble for authors who have utilized the company’s AI for their work. For now, artists who wish to protect their work with copyright will still have to create the expressive work themselves.

Ryan Bickett is a Second Year Law Student at the Benjamin N. Cardozo School of Law and a Staff Editor on the Cardozo Arts & Entertainment Law Journal. Ryan has an interest in Film & TV and Copyright Law.