In November 2024, Coca-Cola released its annual holiday commercial, “The Holiday Magic is Coming.” The ad, a remake of the 1995 Coca-Cola advertisement “Holidays Are Coming,” sparked controversy as it was entirely made by the artificial intelligence (AI) platform, Real Magic AI. At the end of the commercial, Coca-Cola copyrights the commercial by depicting a screen reading, in part, “© 2024 The Coca-Cola Company.” But can an entity really copyright the work that is entirely created by AI?
Generative AI
AI impacts nearly every sector of society, from the military to the legal and medical professions. Generative AI has transformed the arts and sciences by redefining how individuals create, develop, and share ideas. Its integration into content creation has reshaped the digital media landscape, making it no longer solely human-made. Authors, artists, and songwriters use generative AI to produce content that conveys their ideas. While these works are undeniably expressive, many legal experts question whether they lack originality and may not merit copyright protection in the same way as those created entirely by human authors.
U.S. Copyright Office Guidance
Persistent expansion in the use of generative AI by commercial entities has raised significant legal concerns over the ownership of non-human/AI generated content. While courts have begun to address related concerns over the ownership of non-human generated content, existing caselaw does not provide a definitive answer as to whether derivative works produced by generative AI may retain copyright protections. In 2018, the Ninth Circuit Court of Appeals in Naruto v. Slater ruled that a primate could not claim standing under the Copyright Act because only humans can be considered “authors,” a requirement for obtaining copyright protection. While authorship of works generated by a non-human sentient being are an interesting legal parallel to the works produced by generative AI, some have suggested that it may be an insufficient parallel when accounting for human involvement/inputs required to produce a work. Until recently, courts relatively silent on this issue, and without a concrete answer from the courts, content creators were left to look to the U.S. Copyright Office’s (USCO) for guidance.
In 2023, the USCO began an initiative to study how AI may interact with existing copyright law. In 2024 the USCO released its first set of guidance “Copyright and Artificial Intelligence Part 1: Digital Replicas.” In January of 2025, the USCO released “Copyright and Artificial Intelligence Part 2: Copyrightability.”
According to the USCO, authorship of copyrightable works belongs to “the person who translates an idea into a fixed, tangible expression entitled to copyright protection.” Community for Creative Non-Violence v. Reid, 490 U.S. 730 (1989). Under the guidance, whether a given work that was developed using AI retains the requisite human authorship necessary to be eligible for copyright protection depends on how AI was used in the content creation process. Under the guidance, for a work to be eligible to receive copyright protections, it must be creative and originate from human thought. If a work is merely the product of user input, the product is the result of algorithmic functions, and copyright protections are not available. Copyright Registration Guidance: Works Containing Material Generated by Artificial Intelligence at *4. The guidance clarifies that generative AI platforms, such as OpenAI, operate by optimizing user inputs, enabling the AI to “revise or expand them” in ways that enhance the quality of the output, often exceeding the user’s initial instructions. Often, AI also allows users to transcend digital modems. The guidance suggests that ChatGPT “automatically generate[s] tailored, detailed prompts for [OpenAI’s text-to-image model] DALL-E 3” may allow the user to transcend digital mediums in a manner that optimizes the output beyond that which the user ever intended. USCO Guidance Part 2 at *6.
The guidance suggests that works that are “AI-assisted” may be copyrightable, copyright eligibility depends on the extent of human contribution to the final work product. In situations where the final product may be significantly modified from the AI product, or where AI is used in a manner such that their use parallels the creation of a derivative work, the user may be considered an author and retain copyright ownership.
Thaler Supports USCO’s Position On Work Solely Generated By AI
On March 18, 2025, the U.S. Circuit Court of Appeals for the District of Columbia applied a similar understanding to the court in Naruto and upheld the lower court and USCO finding that human authorship is required when seeking copyright protection. The court, in Thaler, stated that “the Copyright Act of 1976 requires all eligible work to be authored in the first instance by a human being.” Thaler v. Perlmutter, No. 23-5233 (D.C. Cir. 2025) at *3. Therefore, works produced solely by AI are not copyrightable.
Conclusion
Although the Guidance provides some clarification, it fails to adequately address the concerns of many legal scholars. In some respects, the USCO guidance is deficient in addressing the impacts of AI on the copyrightability of derivative works. While Coca-Cola’s copyright of their advertisement may be permissible under the recent guidance promulgated by USCO (depending on whether it disclaims the AI generated components of the work), the guidance calls into question the copyrightability of much of the content created by generative AI. Coca-Cola already retained ownership of the copyright for the original content for which “The Holiday Magic is Coming” commercial is a derivative. In situations where songwriters/musicians use generative-AI to develop their works, they run the risk of (1) infringing on others works – as many AI platforms use machine learning to train their algorithms and incorporate protected works into their products or (2) may not be eligible for protection. Thaler’s holding is equally silent on the question of derivative works that bear components created by humans. With many questions still left unanswered regarding the implications of derivative works created from machine learning software, content creators are left waiting for Part 3 of the USCO guidance to be released.
Andrew Bialek is an Associate Editor on the Michigan Technology Law Review.