A few months ago I shared a ruling from a specialized Mexican court that denied authorship rights to AI-generated works. I called it an absurd decision, not because the outcome was wrong, but because the reasoning left too many questions unanswered. Now a significant comparative data point has arrived: the U.S. Court of Appeals for the D.C. Circuit has decided the Thaler case.
Stephen Thaler sought to register copyright in an image titled “A Recent Entrance to Paradise,” naming his “Creativity Machine” system as the sole author. The judges were unequivocal: the Copyright Act requires that the author be a human being. No human author, no copyright.
What the decision holds, and what it does not
The ruling is more nuanced than the headline suggests. The court did not hold that AI-assisted works lack protection. It held that a work entirely generated by AI, without demonstrable human creative input, fails to satisfy the Copyright Act’s originality requirement.
The distinction matters. In practice, the category of “works entirely produced by AI” is nearly empty: behind any AI output lies a human process of selection, instruction, and decision-making. The real issue is not the total absence of human creativity but its insufficiency to reach the required originality threshold; intellectual property law already had the tools to address that.
Consistency with Feist
The court grounds its reasoning in Feist Publications v. Rural Telephone Service, in which the Supreme Court established that effort or investment in resources alone does not justify copyright protection. Protection depends on human originality and creativity, not mechanical labor.
This directly answers the most common corporate argument: “We invested millions in training this model; it deserves protection.” Feist responds that effort is not the criterion. What matters is whether a human author produced an original expression.
Three conditions for protecting AI-assisted works
What the decision implicitly confirms is that AI-assisted creations can receive copyright protection when three conditions are met.
First, the AI must be used as a tool within an identifiable human creative process. Second, demonstrable human creative intent must exist: selecting, discarding, combining, transforming. Third, the copyright must be registered in the name of the human author, not the system.
Under this framework, a lawyer who uses AI to draft a customized contract template, a designer who generates variants and then selects and edits them, or an academic who uses AI to structure an argument that they then develop: all retain rights in their work. The tool does not displace the author; it assists them.
A note on academic integrity
This is precisely the framework I use with my students when discussing academic integrity: we can claim authorship of an AI-assisted text as long as a real creative process took place on our part. If that process did not occur, meaning if the text was generated without genuine editorial decision-making, it must be cited as any other external source.
The analogy with photography is not new but remains accurate. Since the camera’s arrival, courts had to decide whether the photographer was the author of their images or whether the machine stripped them of that status. The answer was clear: the camera is a tool, the photographer is the author. An AI system, however sophisticated, occupies the same position in that chain.
Practical implications
For practitioners working in intellectual property, communications, or technology, the operational implication is direct: documenting the human creative process becomes as important as the output itself. What instructions were given to the system? What selections were made among the generated alternatives? What edits did the author introduce? That documentation is what grounds the legal claim.
What the Thaler decision closes is the path to registering the output of an autonomous system, without any identifiable human input, as one’s own work. What it leaves open is everything else, which is, in practice, the greater part of what we do when we use these tools.
This analysis connects to other lines of inquiry in the Ratio: AI in the Mexican judicial system and liability when AI systems communicate on behalf of an organization.