Terms of Use (OpenAI)
- Document
- undated document
- Event
- no single event
- Retrieved
- 16 September 2026
Start here
Someone who asks an assistant to draft an email, a story or code has two questions tangled together: does the vendor let them use and keep what comes out, and does the law consider them the author of it. The two have different answers, and conflating them is where confusion starts.
What the documents say
OpenAI's Terms of Use, retrieved 16 September 2026, state ‘you retain your ownership rights in Input and… own the Output,’ and that OpenAI assigns ‘all our right, title, and interest, if any, in and to Output’ to the user. Anthropic's Consumer Terms, retrieved the same date, use near-identical language: ‘we assign to you all of our right, title, and interest — if any — in Outputs,’ conditioned on the user's compliance with the terms. Both are contractual promises about the vendor's own claim, not a ruling on copyright law. The US Copyright Office's report on copyrightability, dated January 2025, addresses the separate legal question directly: ‘the Office concludes that, 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.’ The report compares a prompt to instructions given to a commissioned artist, noting that a hiring party who merely describes what they want, without controlling how it is executed, has not traditionally been treated as the work's author under copyright case law.
Check this
Separate the two questions before relying on either: ‘can I use this’ is a contract question the terms answer, in both cases, with yes; ‘do I hold copyright in this’ is a legal question the Copyright Office answers, given current technology, mostly with no, if a prompt is all you contributed. The mechanism the report names — control over the actual execution, not just the request — is the test worth applying to your own work.
What holds and what fails
What holds is that both assignment clauses give you practical permission to use and publish what the tool produces for you, without needing to sort out authorship first. What fails is the assumption that ‘the vendor gave it to me’ means ‘I own the copyright’ — the report is explicit that a vendor cannot assign a copyright interest it does not have, because purely AI-generated material may not be copyrightable by anyone. It does allow that a human's own creative selection, arrangement or substantial editing of AI output can support a copyright claim in that contribution. This is the report's conclusion as of January 2025; it states the Office will keep monitoring the technology and may revisit it.
- Check the assignment language in the terms of whatever tool you use before assuming ownership.
- If copyright matters to your work, add your own substantial editing or arrangement, not just a prompt.
- Do not treat a vendor's usage terms as a copyright determination; they are separate documents.
A vendor's terms and a government report answer different questions: you may be free to use and sell what an assistant wrote for you, while holding no copyright in the parts a prompt alone produced.
Sources & reading trail
States the user owns Output and retains ownership of Input, with OpenAI assigning its rights in Output to the user.
Source published: Not established · Retrieved: 16 September 2026
States Anthropic assigns its right, title and interest in Outputs to the user, subject to compliance with the terms.
Source published: Not established · Retrieved: 16 September 2026
Concludes that, given current technology, prompts alone do not provide sufficient human control to make a user the legal author of AI output.
Source published: 1 January 2025 · Retrieved: 16 September 2026
Documentation, regulator guidance and studies establish the record; the checks and the boundary are AI Use Field Guide editorial analysis. This retrospective draft does not imply the site published on the event date.