The Death of Art(ificial)istry

By Abigail Kuhlman. 

Art v. Artificial

In August of 2022, Jason Allen nervously entered his first fine arts competition under the “Digital Arts/Digitally-Manipulated Photography” category at the Colorado State Fair. Allen’s submission, an image titled “Théâtre D’opéra Spatial” (“Space Opera Theater” in English), depicted three classical figures wearing billowing, regal dresses and standing in a futuristic, Baroque-style opera hall; the figures are faced away from the viewer, looking through a glowing, almost portal-like viewport and out onto a beautiful landscape. To Allen’s surprise and excitement, the judges were enraptured by this image and awarded it first place. But when Allen took to Twitter (now X) to share his exciting win, it wasn’t messages of congratulations that he would receive. Upon revealing he had created his award-winning piece using Midjourney, an artificial intelligence (“AI”) program that transforms written prompts into images, Allen found himself embroiled in intense controversy. He drew sharp criticism from artists and art enthusiasts across the web, with one user’s Tweet, now since removed, painting Allen’s win as the “death of artistry.” 

Since Allen’s win, AI-generated works have continued to beat out their human-created counterparts, sparking similar controversy and disapproval amongst the art community. At its core, this disapproval reflects a budding legal issue within copyright law: are the artistic products of generative-AI art? The generated works are—and several Colorado art judges would agree—certainly satisfying to viewers. Moreover, had its AI-origins been unknown, even the harshest of art critics would characterize the work as “art” when looking at the product alone. Still, critics argue that pressing a couple of buttons on a machine and instantly churning out an award-winning masterpiece is the product of the AI’s creation, not the user. This characterization, however, is somewhat oversimplified. While it is true that generative-AI programs, such as Midjourney, can produce detailed images from written prompts in a matter of seconds, creating works consistent with the user’s vision requires investing significant time and effort into the creative process. Allen, for example, spent over eighty (80) hours on “Théâtre D’opéra Spatial” brainstorming how to phrase the written prompt and tweaking it to adjust artistic elements such as lighting and color harmony. By the time Allen was finished, approximately 900 different drafts had been made. Some argue that this substantial degree of human input and control indicates that the human is the author of the generated work. Therefore, while AI programs may swap paintbrushes for prompts, an artist’s use of AI is no different than how a photographer uses a camera: as an artistic tool.

As AI has continued to rapidly develop, similar controversies have proliferated within the art world. Despite this division, the uptake of AI programs by the artistic community has become increasingly more commonplace and has fundamentally changed how artists approach creativity and innovation. While some artists express concern that AI could replace human creativity, for many artists, AI has become (and remains) an indispensable tool to enhance their artistic capabilities, develop new styles and techniques, streamline the creative process, and expand their imagination and the possibilities for artistic creation. Over the past several years, applicants have begun submitting AI-generated works for registration by the Copyright Office; however, most of these applications have been denied. In response to these developments, the Copyright Office concluded that public guidance was needed on AI-generated works.

Copyright Office–Statement of Policy

On March 16, 2023, the Office issued a Statement of Policy clarifying its practices and providing practical guidance to authors regarding the registration of works containing AI-generated materials. The Statement began by reiterating the “well-established” copyright principle requiring creative works to be the product of human authorship. Accordingly, any creative works whose expressive elements originate from non-humans—including divine beings, monkeys, and now AI—will not be registered by the Copyright Office. However, in some circumstances, works containing AI-generated materials can have sufficient human authorship to be copyrighted. For example, if a human selects and arranges the generated materials in such a way that it qualifies as a “compilation” or modifies the generated material to the extent that it qualifies as a “derivative work,” then those human-authored aspects can be protected—so long as the AI-generated material itself remains unprotected. 

Additionally, the Statement does contemplate the possibility of registering AI-assisted creative works. The precise line between AI-generated materials and human creation has yet to be defined, but important factors to consider moving forward will include: (1) how the AI programs operate and are used; (2) whether the AI program is merely an assistive instrument; (3) whether the user exercises “ultimate creative control” over how the AI interprets prompts and generates material; (4) whether the work’s traditional elements of authorship (i.e., the literary, artistic, or musical expression or elements of selection, arrangement, etc.) are conceived and created by a human; and (5) for works containing AI-generated materials, whether the AI’s contributions are the product of “mechanical reproduction” or by a human’s “own original mental conception, to which [the author] gave visible form.” Other factors, such as whether the written prompts received by the AI program are sufficiently creative for copyright protection and whether the AI program allows users to provide “feedback” by adding or revising prompts, will not be determinative.

The Statement also provides authors with practical guidance on submitting and updating registration applications for works containing AI-generated materials. In general, authors have a duty to disclose AI-generated materials and explain the human author’s contributions to the work, unless the generated materials amount to a de minimis use (e.g., brief quotes, short phrases, etc.). If authors are unsure whether a use is de minimis, a general statement that the work contains AI-generated materials will be sufficient and the Office, if necessary, will follow up with the author when the application is under review. For pending applications, authors must take steps to verify that AI-generated materials have been adequately disclosed; if not, authors must report that omission to the Office. For a work that has already been registered, authors must correct the public record by submitting a supplementary registration that discloses the inclusion of AI-generated works. Failing to do so could result in a cancellation of the author’s registration.

An Unclear Future for Creative Industries

While the Statement definitively forecloses the possibility of protecting AI-generated works without human authorship, other issues related to the development and use of AI have yet to be addressed (e.g., use of copyrighted works to train AI and how those outputs should be treated, infringement, etc.). Recognizing this, the Office has launched an agency-wide AI Initiative and published a Notice of Inquiry seeking public input on the legal and policy issues raised by generative AI. In particular, the Notice will address “(1) the use of copyrighted works to train AI models; (2) the copyrightability of material generated using AI systems [including whether there may be circumstances in which users exert sufficient control over the AI such that the generated work is human authored]; (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.” The information gathered from these initiatives will be used to inform the Office’s regulatory work and advise Congress on potential areas for congressional action.

The Office’s actions are a positive indication of its commitment to stay involved in the developing legal landscape of AI and its implications on copyright law. However, until formal policies are established—and assuming advancements in AI haven’t “outgrown” these policies before they’re implemented—artists, businesses, and developers already developing and using AI will be blindly navigating this regulatory void. For Arizona businesses and creators, this is especially concerning. As a growing hub for business, innovation, and technology, survival in Arizona’s competitive climate pressures businesses and creators to adopt AI or risk being left behind. Moving forward, artists, businesses, and developers will need to pay close attention to any legal developments in AI and establish risk-mitigation strategies and ethical policies to guide themselves through these uncharted waters.

"Théâtre D'opéra Spatial" by Midjourney using a prompt by Jason M. Allen (2022).