The Copyright Office says a prompt alone is not authorship

The test

The U.S. Copyright Office published 'Copyright and Artificial Intelligence, Part 2: Copyrightability' on 29 January 2025, per the Office's own AI report index page, the second installment of a multi-part report examining how existing copyright law applies to AI-related questions, directly relevant to whether an AI-assisted short-form video can be registered.

What the evidence says

The report states its central conclusion plainly: 'given current generally available technology, prompts alone do not provide sufficient human control to make users of an AI system the authors of the output,' because prompts function as instructions conveying 'unprotectible ideas' rather than control over how a system executes them -- even highly detailed prompts, the report says, do not yet control the generation process itself. The report draws a more permissive line for editing tools: features that let a user 'control the selection and placement of individual creative elements,' such as region-based editing and canvas-style revision tools, can produce a copyrightable output 'in those cases where they do' meet the ordinary originality threshold, decided case by case. A third point covers mixed works: including AI-generated material inside a larger human-authored work 'does not affect the copyrightability of the larger human-authored work as a whole,' the report says, giving the example of a film with AI-generated effects or background art that remains copyrightable even though those AI elements alone are not. The report builds on the Office's 2023 AI Registration Guidance, which already required applicants to disclose AI-generated material and describe the human author's own contribution.

The sample and the variance

This is the Copyright Office's own report and reflects the Office's registration practice and legal analysis as of January 2025, not a court ruling binding every future case, and it explicitly addresses copyrightability only. It does not set platform monetization policy, which remains each platform's own separate rule.

What to try next

As an editorial reading of the report rather than legal advice, a short-form producer using generative AI tools can treat heavy post-generation editing -- selecting, arranging and revising individual elements -- as the documented path toward a registrable claim, and treat a prompt-only output as the documented path that currently is not one.

• Did a human select, arrange or revise individual generated elements, or only submit a prompt and accept the first output?
• Is the AI-generated material one component inside a larger human-authored work, or the entire claimed work itself?
• Has a court applied this report's reasoning to a specific dispute since January 2025?

The report does not forbid using generative AI in a short-form production. It draws a line, backed by the Office's own stated reasoning, between prompting a system and controlling one.