THE HIDDEN BATTLE BEHIND AI PROMPTS
ABSTRACT
Each use of tools like ChatGPT or Midjourney raises legal questions about ownership rights over generated content. This includes inquiries about whether the initial input can be protected material and how the AI’s response is categorized under existing copyright laws. As AI increasingly influences content creation, traditional copyright frameworks face challenges in addressing these novel issues. The discussion focuses on whether users can claim ownership of their inputs and the broader implications for users reliant on prompts.
INTRODUCTION
Copyright law traditionally protects original works created by humans, but the rise of generative AI complicates this framework. As users provide prompts AI to generate text or imagery, the distinction between human authorship and machine output becomes unclear. This raises legal questions regarding the copyright status of such works and who would hold that copyright. The blog examines current legal perspectives on these issues.
THE BASIC PROBLEM: IDEAS VERSUS EXPRESSION
What you think up can’t be copyrighted, only how you actually write it down counts. Right now, people are arguing hard about where that line should sit with AI stuff. Back in January 2025, the U.S. Copyright Office said most prompts act like basic directions, more like raw concepts than real creative work. Since these inputs don’t shape exactly how the tool builds a response, they lack the kind of hands-on touch needed under copyright rules. That means whoever types the prompt isn’t seen as the creator of what comes out the other end.
A robe-clad cat puffing on a pipe appeared when The Office tested an open-ended description. Without clear directions about fur shade, posture, or ancestry, the machine leaned into its learned patterns. What emerged carried the software’s decisions more than the writer’s intent – even if the starting idea felt imaginative.
CAN A PROMPT ITSELF BE COPYRIGHTED?
This discussion is separate from the question of whether the user’s input in the prompt can be considered as a copyrighted literary work. On this point, the Copyright Office has not ruled out the possibility. A prompt that is detailed and creatively composed, conveying a distinctive artistic idea through rich, expressive language rather than a simple, general instruction, might contain original expression eligible for copyright, independent of the AI’s resulting output. For instance, a brief command such as “paint a picture of a dragon” probably wouldn’t qualify, but a lengthy, imaginative description of a surreal cityscape could satisfy the standard for originality required by copyright law.
Still, this distinction provides limited practical benefit. Even if the prompt is protected as an original piece of writing, that protection does not cover the image or content the AI produces in response to it.
THE THALER LITIGATION: A COURT WEIGHS IN
The most definitive legal ruling to date comes from Thaler v. Perlmutter, in which Stephen Thaler attempted to register an artwork created entirely by his AI system, naming the system as the sole author. The D.C. Circuit upheld the refusal, stating that under the Copyright Act, only works originally authored by a human qualify for protection, though it emphasized this does not rule out copyright for works developed with AI assistance. The Supreme Court subsequently declined to review the decision.
A more pertinent case, Allen v. Perlmutter, now challenges how to define authorship for AI-generated works based on user input. Artist Jason Allen is contesting the denial of copyright registration for an image produced using more than six hundred carefully crafted Midjourney prompts. His appeal brings into focus a key unresolved issue: whether extensive and detailed prompting can ever constitute sufficient creative contribution to qualify as authorship.
WHY THIS MATTERS BEYOND THE COURTROOM
The practical implications are significant. Companies using AI-generated content for marketing or design purposes may discover that such content lacks exclusivity, making it vulnerable to replication by competitors. Most legal systems, including India’s under the Copyright Act of 1957, require human authorship for copyright protection, and there is currently no specific legal framework addressing works created with AI assistance. For individual creators, this creates a pressing dilemma: how much modification or input is necessary to transform AI-assisted output into something eligible for protection? In the absence of clear regulations, the best approach is to meticulously record the creative process and ensure that any AI-generated material is substantially transformed through original human contribution before being claimed as a protected work.
CONCLUSION
The issue of whether prompts can be copyrighted lacks a definitive answer. While a prompt may qualify for copyright if sufficiently creative, this does not extend to the AI-generated output, which is only protectable when human creative choices are evident. Thaler has established that machines cannot be authors, leaving unclear the extent of human control required to claim authorship over machine outputs. As generative AI tools evolve, this question is likely to resurface, particularly in cases like Allen v. Perlmutter, highlighting the ongoing negotiation of copyright values—originality, human creativity, and authorship—triggered by AI prompts.
REFERENCES
Thaler v. Perlmutter (2025) 130 F.4th 1039 (D.C. Cir.)
Thaler v. Perlmutter (2023) 687 F. Supp. 3d 140 (D.D.C.)
Burrow-Giles Lithographic Co. v. Sarony (1884) 111 U.S. 53
The Copyright Act, 1976 (United States)
The Copyright Act, 1957 (India)
U.S. Copyright Office, Copyright and Artificial Intelligence, Part 2: Copyrightability (2025), [https://www.copyright.gov/ai/]
U.S. Copyright Office, ‘NewsNet Issue 1060: Copyright Office Releases Part 2 of Artificial Intelligence Report’ (29 Jan 2025), [https://www.copyright.gov/newsnet/2025/1060.html]
Congressional Research Service, ‘Generative Artificial Intelligence and Copyright Law’ (Library of Congress, 2025), [https://www.congress.gov/crs-product/LSB10922]