The Supreme Court’s recent denial of certiorari in Thaler v. Perlmutter[1] leaves undisturbed the decision from the DC Circuit’s ruling that the Copyright Act requires human authorship as a prerequisite for registrability. By declining review, the Court preserved the consensus that AI‑generated works created without meaningful human creative contribution are not eligible for copyright protection.   

But that consensus is not new. Modern AI‑authorship decisions rest on a foundation laid more than a century ago, when the Supreme Court confronted a strikingly similar question about whether a machine–aided technology — photography — could threaten the definition of authorship.

The Enduring Logic of Burrow‑Giles

In Burrow‑Giles Lithographic Co. v. Sarony[2], the Court rejected the argument that photographs were uncopyrightable because they were produced by a machine rather than a human hand. Justice Miller held that copyright protects only “original intellectual conceptions of the author.”[3] Purely mechanical output showing “no originality of thought,” he warned, fell outside copyright’s scope.[4] 

Critically, the Court did not stop at that formulation. It catalogued the specific creative choices that made Sarony (not the camera) the author: he posed Oscar Wilde, selected and arranged props, controlled lighting and shading, elicited the subject’s expression, and produced a harmonious composition reflecting his own mental conception. Later cases reaffirm that protectable authorship turns on these kinds of human creative judgments.

While Burrow-Giles does not contain a freestanding declaration that human authorship is required (the Court wasn’t presented with that precise issue), it defined authorship so thoroughly in terms of human intellectual agency that the human authorship requirement is a logical extension of that definition, not a departure from it. This formulation leaves little room for a machine operating without human direction or conception. 

Thaler Was Never a Close Case

Over 140 years later, the same framework disposed of Dr. Stephen Thaler’s copyright claim. Dr. Thaler, a physicist and founder of Imagination Engines, Inc., who holds several AI patents and is an advocate for the legal recognition of machine creativity, filed what many observers view as a deliberately orchestrated series of test cases across patent and copyright law.

At the center of the copyright effort was the piece “A Recent Entrance to Paradise,” generated, according to Thaler, independently by his DABUS (Device for the Autonomous Bootstrapping of Unified Sentience) AI system. The Copyright Office summarily rejected the application because, by Thaler’s own admission, no human contributed to its expressive elements. The district court and the DC Circuit agreed, with the appellate court holding that a copyrightable work must be “authored in the first instance by a human being.”[5] Throughout the litigation, the Copyright Office framed the structural argument plainly and convincingly: among other provisions, the Copyright Act ties protection to the author's life plus seventy years, a framework that cannot be applied to machines, which have no legally cognizable lifespan. 

By denying certiorari, the Supreme Court signaled it sees no doctrinal uncertainty worth resolving (for now). The consensus stands: works created by AI authorship, alone, are off the table for copyright protection under current law. 

What Remains Unclear: Human Contribution in AI‑Assisted Works

While autonomous AI authorship is settled, the boundary between “AI‑assisted” and “AI‑generated” is anything but. Thaler never determined how much human involvement is necessary to support a copyright claim for AI-assisted works. So when are detailed prompt engineering and human input, akin to Sarony’s photograph, sufficient to convey human authorship? For now, that answer turns on the specific claims applicants make in an application. 

The Copyright Office has drawn a clear line for prompt-only authorship claims: “Based on the functioning of current generally available technology, prompts do not alone provide sufficient control [to warrant authorship].”[6] However, in the same 2025 report, the Office recognizes that “neither the use of AI as an assistive tool nor the incorporation of AI-generated content into a larger copyrightable work affects the availability of copyright protection for the work as a whole.” 

With these signposts in mind, the Office has accepted an application where human authorship exists in the selection, arrangement, and modification of AI-generated material. In its Zarya of the Dawn[7] decision, the Office canceled the registration covering Midjourney-generated comic book artwork but reissued the registration covering only the book’s human-authored text as well as the selection, coordination, and arrangement of its written and visual elements.

However, the same year, the Copyright Appeal Board affirmed a refusal to register artwork that was created by taking an original, human-authored photograph, feeding it into an AI-painting app as the base image, inputting Van Gogh’s The Starry Night as the style image, and selecting a value for the amount of style transfer and image interpolation. The Board found that it was the AI-app, not the human applicant, that was “responsible for determining how to interpolate the images [to produce the output].”[8]

In a registration refusal appeal that is currently pending in the District of Colorado,[9] content creator Jason Allen argues that his application for Theatre D'Opéra Spatial, a Midjourney-assisted artwork that was the product of Allen’s 624 detailed, iterative Midjourney prompts, should not have been refused. Notably absent from other refusal decisions is the fact that Allen edited the original Midjourney image with Adobe Photoshop and upscaled it using a tool called Gigapixel AI to create the final image. 

Most recently, the Office granted registration for A Single Piece of American Cheese, artwork whose initial image was created by AI from a text prompt but then edited by the human user by a process called “inpainting” to produce the final image. Observing that the human input resulted in non-trivial changes to the original AI-generated image, the Office concluded that the final image contained a sufficient amount of human original authorship in the selection, arrangement, and coordination of the AI-generated material that may be regarded as copyrightable.

The registration decision for A Single Piece of American Cheese hardly clarifies the issue of what constitutes “non-trivial changes” to original AI-generated images. All eyes are on Allen as it may become the first case to articulate a judicial standard for when human contribution is sufficient to constitute human authorship for AI-assisted works. This much is clear, though: meaningful human creative contribution is a prerequisite for copyright registrability of AI-assisted works. 

What Businesses Should Know

For companies integrating generative AI into creative workflows, several practical principles are already clear:

1. Document human creative control at every stage — prompt selection, arrangement, refinement, and editing of generative outputs. An evidentiary record of human input that leads to a final work is crucial to demonstrate human authorship. 

2. Distinguish clearly between AI‑assisted and AI‑generated content. Only the former is protectable. Simply prompting and clicking “generate” will not lead to copyrightable works. 

3. Audit creative workflows to pinpoint organizational AI use and keep records of where human creativity enters the chain of production across all domains. AI-generated source code, marketing materials, and other AI-generated assets could lead to copyrightability issues.