In 2023, the US Copyright Office launched an initiative to study the intersection of copyright law and AI. Drawing on feedback from a wide array of stakeholders such as AI developers, lawyers, creatives, and academics, the USCO is in the process of publishing a three-part report on the subject. Each part focuses on a different subtopic This post will examine copyrightability, with special emphasis on the role of human creators, the distinction between creative and assistive uses of AI, and the role of prompts.
What is copyright?
American copyright law is founded on Article I, clause 8, section 8 of the United States’ Constitution, which states that Congress can: “promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries.”
In a nutshell, copyright is a legal concept that gives creators the sole right to use and distribute their original works for a limited period of time. It’s meant to incentivize creativity by ensuring that creators can benefit financially from their works. Otherwise, it might be hard for them to justify the time, energy, and money spent on creating.
Enter the robots
Technology and creativity have been intertwined since our Stone Age ancestors painted the walls of caves with pigments they’d fashioned from their environment. Of course, modern technology has added new variables to the mix since it can have a profound impact on the resulting work. As the USCO’s report points out, there is nothing new about this debate. As far back as 1965, the Register of Copyrights wrote that:
“The crucial question appears to be whether the “work” is basically one of human authorship, with the computer merely being an assisting instrument, or whether the traditional elements of authorship in the work (literary, artistic, or musical expression or elements of selection, arrangement, etc.) were actually conceived and executed not by man but by a machine.”
This led the USCO to decide that it wouldn’t be prudent to deny copyright registration to a work simply because a computer was involved in its creation. Instead, each registration request would need to be decided on a case-by-case basis. The USCO applies these same principles to work involving AI:
“Where AI merely assists an author in the creative process, its use does not change the copyrightability of the output. At the other extreme, if content is entirely generated by AI, it cannot be protected by copyright.”
The workings of generative AI
The USCO report notes that generative AIs usually work by having the user input a prompt that can be either simple (“a cat dancing”) or complex (“a cat flamenco dancing in a bright red dress painted in the style of Renoir”).
But what you see in the prompt isn’t necessarily what you’ll get, and the AI may add or subtract elements with no apparent rhyme or reason. In the words of the USCO,
“[a]lthough AI technology continues to advance, uncertainty around how a particular prompt or other input will influence the output may be inherent in complex AI systems built on models with billions of parameters.”
While some generative AIs such as the image-generator Midjourney allow users to set a seed that can add a measure of consistency to the results so that reusing a prompt produces similar images, the results aren’t guaranteed.
Only humans can be authors…
As things stand, only humans can be recognized as the author of a work for copyright purposes. This is supported by case law. In Urantia Foundation v. Maaherra, the Court of Appeals for the Ninth Circuit held that text purportedly created by “non-human spiritual beings” was ineligible for copyright. In Naruto v. Slater, the same court also found that a photo taken by a monkey was also ineligible for copyright (“we conclude that this monkey—and all animals, since they are not human—lacks statutory standing under the Copyright Act”).
However, if an AI is ever deemed to be sentient, it might change the calculus entirely. Although that claim has already been made (and rejected) in a copyright case, the ever evolving technological landscape may force a rethink before too long.
…but technology can help them create
While the outcome in those cases might seem obvious, the decision isn’t always as clear-cut. In the late 19th century, the Supreme Court had to contend with the question of whether or not photographs were copyrightable in the case of Burrow-Giles Lithographic Co. v. Sarony. Although the defendant argued that photographs were not copyrightable since they were entirely produced by a machine (i.e., the camera), the Court rejected this approach.
The justices defined an ‘author’ as “he to whom anything owes its origin; originator; maker; one who completes a work of science or literature.” They also noted that the human photographer plays a key role in creating the photograph since they may pose the subject, incorporate costumes or props, or adjust the lighting to create different effects. So while the use of technology does not automatically preclude copyright registration, the work must include some element of human ingenuity.
However, not all contributions have the same weight in the eyes of the law. In Community for Creative Non-Violence v. Reid, the Supreme Court held that the sculptor who created a statue was the author and not the non-profit that commissioned it even if they happened to provide detailed feedback.
When applied to AI, the USCO believes that “there is an important distinction between using AI as a tool to assist in the creation of works and using AI as a stand-in for human creativity.” For example, if someone puts a prompt into a generative AI and then references, but doesn’t incorporate, the AI’s output into their work, the use of AI shouldn’t affect its copyrightability.
These basic principles seem clear cut in theory, but in practice they can be quite messy.
Does a prompt make an author? No, but…
The USCO argues that, at the moment, simply inputting a prompt alone isn’t enough to make someone the author of an AI-generated work. “Prompts essentially function as instructions that convey unprotectible ideas. While highly detailed prompts could contain the user’s desired expressive elements, at present they do not control how the AI system processes them in generating the output.”
The USCO notes that there is always a gap between the prompt and the resulting output. “In other words, prompts may reflect a user’s mental conception or idea, but they do not control the way that idea is expressed.”
To make their point, USCO staffers used Google’s Gemini AI to create an image of a cat
Although they provided a detailed prompt, the resulting image didn’t fully adhere to the instructions. The cat wears glasses, smokes a pipe, and reads a newspaper, but it’s not in a “highly detailed wood environment.” Gemini also added elements, like having an “incongruous human hand” hold the newspaper.
This disconnect between the prompt's instructions and the final output illustrates the USCO's point that users cannot precisely control AI expression through prompts alone, undermining claims of authorship based solely on prompt writing.
Submitting revised prompts isn’t enough to qualify for authorship. The USCO likens this to re-rolling dice. “No matter how many times a prompt is revised and resubmitted, the final output reflects the user’s acceptance of the AI system’s interpretation, rather than authorship of the expression it contains.” However, if and when the technology evolves to the point that humans can have more control over the “expressive elements,” a different approach may be warranted.
Expressive prompts are a different story
Not all prompts are alike. Some generative AIs allow the user to input original creations such as photos or illustrations with an instruction to the AI to modify it in some way. For example, an artist drew an illustration and then inputted it to a generative AI along with a text-based prompt.
When applying for copyright, the artist explicitly disclaimed “any non-human expression” in it such as the more realistic depiction of some of the facial features and the flowers. The USCO deemed this copyrightable, though it noted that registration was “limited to unaltered human pictorial authorship that is clearly perceptible in the deposit and separable from the non-human expression that is excluded from the claim.”
But at the same time, the USCO refused to register an image created using a generative AI that could make images resemble Vincent van Gogh’s The Starry Night. Although the resulting image was based on a photograph supplied by the applicant, the USCO found that the end result did not “contain sufficient human authorship necessary to sustain a claim to copyright.” The image provided by the applicant wasn’t clearly discernible in the output.
Modifying or arranging AI-generated work
The USCO argues that a human author should be able to claim copyright when they have modified or arranged AI-generated work in a creative way even if the AI-generated work itself isn’t necessarily copyrightable.
The report highlights Midjourney’s “Vary Region and Remix Prompting” feature which allows more granular control over the selection, arrangement, and content of the final output. While works created with such tools can be copyrightable, each decision must be made on a case-by-case basis.
Should AI-generated content receive special treatment?
The USCO report rejects the notion that the law should be changed to give AI-generated content special treatment. “Because copyright requires human authorship, copyright law cannot be the basis of protection for works that do not satisfy that requirement.” They also felt that existing law provided sufficient protections for AI-generated content. Interestingly, they also mentioned that, if AI is too incentivized, it could have a negative impact on the expression of human creativity.
“We share the concerns expressed about the impact of AI-generated material on human authors and the value that their creative expression provides to society. If a flood of easily and rapidly AI-generated content drowns out humanauthored works in the marketplace, additional legal protection would undermine rather than advance the goals of the copyright system. The availability of vastly more works to choose from could actually make it harder to find inspiring or enlightening content.”
The report notes that many creators have already seen their ability to profit from their work being negatively affected by AI. “If authors cannot make a living from their craft, they are likely to produce fewer works. And in our view, society would be poorer if the sparks of human creativity become fewer or dimmer.”
Challenges and opportunities
The evolving landscape of AI and copyright law presents both opportunities and challenges. While technology can significantly enhance human creativity, the USCO believes it’s essential to maintain the core principle that copyright protection is fundamentally rooted in human authorship. Striking a balance between embracing technological advancements and preserving the value of human ingenuity can create a copyright system that supports innovation while protecting the rights and contributions of human creators. The ongoing dialogue and forthcoming reports from the USCO will undoubtedly play a crucial role in shaping the future of copyright law in the age of AI.
The first part of this series dealing with digital replicas can be found here.
About the author,
José Mauricio Duque
