Reconciling Copyright Originality for Photography and Generative Artificial Intelligence

Abstract

Copyright protection requires creative choices—creative choices that manifest in the work for which protection is sought. We all understand this point in theory, but in practice it has proven difficult to apply to visual works. The contemporary treatment of photography and generative artificial intelligence (AI), two different technologies used to create images, illustrates this. Examples of courts holding photographs unprotected by copyright are few and far between; in contrast, the Copyright Office and federal courts have been reluctant to grant copyright protection to AI-generated elements of visual works. To sharpen the contrast with real-world examples: A photo of a sudden arrest, snapped without thought on a smartphone with default settings, has been found to be protected by copyright, while an AI-generated image refined over hundreds of prompts to illustrate a graphic novel has been denied protection. These seemingly incongruous outcomes may pose a challenge for copyright law as it seeks to regulate intellectual property rights across different media and technologies.

We do not argue here that the Copyright Office has been too harsh on a controversial emerging technology. Rather, we suggest that contemporary treatment of photography has become lax, prone to granting copyright with little or no analysis. Often, courts assume any photograph is protected by copyright and defer examination of what elements in the photograph (if any) might be protectable until forced to do so in connection with the infringement or fair use analysis. Instead, we suggest, courts should fully examine the originality of a work at the copyrightability stage. Recent Copyright Office actions on generative AI can provide a useful roadmap for a more rigorous originality analysis of visual works—specifically, by emphasizing the importance of tracing purported creative choices to an expressive result that manifests in the final work. Existing doctrines designed for infringement analysis, such as “thin copyright” and “dissection,” also offer useful conceptual models. Just as not every element in an image will be deemed creative for purposes of determining infringement, not every human choice should necessarily be deemed creative for purposes of finding originality. By closely analyzing the connection between creative choice and resulting expression in photographs at the originality stage, courts can return coherence and rigor to copyright doctrine for visual works.