Stephen Thaler spent years trying to register a copyright for an image
produced by his DABUS AI system. He argued the machine was the author.
The Copyright Office said no. The DC Circuit said no. On March 2, 2026,
the Supreme Court declined to hear his appeal in Case No. 25-449,
and with that, the question of whether an AI can be a copyright author
under current US statute is settled.
It cannot.
What the Ruling Actually Says
The DC Circuit’s ruling, now final, held that the Copyright Act requires
human authorship. The Copyright Office’s longstanding position, that
registration requires “an author who is a human being”, is consistent
with both the statute’s text and its legislative history.
As IP/Decode noted in its coverage of the cert denial, the DC Circuit
applied this principle directly to the facts Thaler presented: a system
operating autonomously, without human creative direction, produced the work.
That is the scenario the ruling forecloses.
What the ruling does not do is define the line precisely. “Human authorship”
is the requirement. What constitutes sufficient human creative control in
AI-assisted workflows, where a human prompts, directs, selects, and modifies
but does not physically generate the work, is not resolved. Legal analysts
interpret the standard as requiring meaningful creative control, but no court
has drawn a bright-line test.
Morgan Lewis’s analysis frames the open question accurately: businesses
working with AI-assisted content can protect works where humans exercise
meaningful creative control, but the precise threshold remains to be developed
through future cases.
A Three-Stage Case History
To understand where things stand, the trajectory matters.
Thaler first challenged the Copyright Office’s refusal to register his
AI-generated image in the district court, arguing the statute didn’t
require human authorship. The district court disagreed, holding human
authorship was required. The DC Circuit affirmed. SCOTUS denial on
March 2 completes the circuit, three consecutive tribunals, same answer.
This wasn’t a close case on the autonomous AI question. The legal analysis
at every stage centered on one fact: no human made the creative choices.
The DC Circuit didn’t need to define what “sufficient” human involvement
looks like because Thaler’s submission didn’t involve any. The court
addressed the easy end of the spectrum. The hard middle remains.
What This Means for Teams Using Generative AI
The practical consequences are different depending on how a team uses AI.
For workflows where AI generates output with minimal human direction,
a prompt that produces a finished image or text, used largely as generated,
the ruling is a clear signal: that output is unlikely to receive copyright
protection under current US law. Anyone building a business model on
proprietary AI-generated creative content should treat that output as
unprotected until further legal development says otherwise.
For workflows where humans use AI as a tool, directing the creative
process, making substantive selection and arrangement decisions,
modifying outputs, combining AI-generated elements with human-authored
structure, the position is more defensible. No court has said that
AI-assisted work with genuine human direction is unprotectable.
The Copyright Office has issued guidance acknowledging that AI-assisted
works with sufficient human authorship may qualify. But “sufficient” is
still undefined.
Three questions your IP counsel should be evaluating now:
First, can you document the human creative decisions made during the
production of your AI-assisted content? Courts will likely look at
the nature and extent of human involvement. Records matter.
Second, does your current content workflow involve enough human direction
to support a human authorship claim? “Prompt engineering” is a genuine
creative skill, but its sufficiency for copyright purposes is untested.
Selection, curation, and substantial modification carry stronger arguments.
Third, what is your competitive risk if key content assets turn out to be
unprotected? If competitors can freely copy AI-generated content you’ve
published, your IP strategy needs to account for that possibility now.
The Road Ahead
The Supreme Court’s silence on the merits doesn’t foreclose future cases.
A different plaintiff, with a different set of facts, one involving more
documented human creative direction, could reach a different result.
Congress could amend the Copyright Act. The Copyright Office could issue
more specific guidance.
None of those developments are on a predictable timeline. For now, the
operating standard is what the DC Circuit established: human authorship
is required, autonomous AI work is excluded, and the middle ground is
developing case by case.
That is the landscape compliance teams, IP counsel, and AI product
leaders are navigating. Thaler v. Perlmutter didn’t create a crisis,
it clarified one side of a line that was already forming. The work now
is building workflows, documentation practices, and IP strategies that
hold up on the side of the line where copyright protection is possible.
*This article involves legal interpretation with direct compliance implications.
Consult qualified IP counsel before making specific copyright protection
decisions about your AI-generated content portfolio. The ruling establishes
a general principle; application to specific works and workflows requires
legal analysis.*