A machine cannot register a copyright it authored alone

The writer's problem

Stephen Thaler applied to register a visual work he said his self-built AI system, the Creativity Machine, produced entirely on its own, listing the machine as author and himself only as the machine's owner. The Copyright Office refused registration, and on 18 August 2023 Judge Beryl Howell of the District Court for the District of Columbia affirmed that refusal in a memorandum opinion published on the Copyright Office's own AI initiative page. The problem the case posed was narrow by design: not how much human input a copyright needs, but whether zero input can ever qualify.

What the documents show

The opinion states that human authorship is an essential element of a valid copyright claim, tracing that requirement through Burrow-Giles Lithographic Co. v. Sarony, the 1884 photography case, and through cases denying copyright to works claimed by non-human or divine authors. Critically, the court's holding rests on the administrative record Thaler himself built: his registration application stated the work was created autonomously with no human involvement, and judicial review of the Copyright Office's decision is limited to that record. When Thaler later argued in court that he had directed and controlled the AI, the opinion states those claims came too late and contradicted what he told the Register during the application process. The opinion explicitly calls this case simple compared with harder questions still open, such as how much human input is enough when an artist uses AI as one tool among others, and it does not purport to answer those questions.

The editorial choice

A writer or artist using AI tools with real creative control over the result is not the subject of this ruling; the opinion itself distinguishes a claim of full machine autonomy from a claim of human-directed use of a tool. Editorial guidance drawn from this case should say only that autonomous, undisclosed machine authorship fails, not that any AI-assisted work is uncopyrightable.

What stays with the author

What the case leaves unresolved is exactly how much human creative control converts an AI-assisted output into a human-authored work, a question the opinion flags as open rather than settled. The Copyright Office's own AI initiative page, which hosts this litigation record alongside the Office's related registration decisions, does not resolve that question either. An author still has to be able to show, and accurately describe on a registration application, what they personally contributed.

• Did the applicant claim any human creative control, or did the record show none, as in this case?
• Is a cited AI-authorship dispute about full machine autonomy or about a human using AI as an assistive tool?
• Does a summary of this ruling overstate it into a general rule about AI-assisted work?

Thaler's case turned on the facts he put in his own application; a different applicant who accurately describes real creative control over an AI-assisted work faces a different, still-open question.