Before You Publish That AI-generated content Answer These 4 Questions

14 min read

TL;DR

• The US Copyright Office concluded in January 2025 that copyright does not extend to purely AI-generated content, and that prompts alone do not give a user enough control over the output to count as authorship.
• Using AI as an assistive tool costs you nothing. The Office draws its line between AI assisting human expression and AI standing in for it.
• Three contributions do count: your own copyrightable work perceptible in the output, creative selection and arrangement of AI generated material, and creative modification of it.
• The DC Circuit affirmed the human authorship requirement in March 2025. The Supreme Court declined to review it in March 2026.
• The UK still protects computer generated works for 50 years under section 9(3) of the Copyright, Designs and Patents Act 1988. A government report published on 18 March 2026 recommended removing it.
• From 2 August 2026, EU transparency rules require generative AI outputs to be marked in a machine readable format, with penalties reaching 15 million euro or 3 percent of worldwide turnover.

Here is a question worth asking about the last thing your business published. If a competitor copied it word for word tomorrow, could you stop them?

For a growing share of AI-generated content, the answer is no, and the reason is not a loophole or an untested theory. Who owns AI-generated content is a settled question at the agency that administers US copyright law, backed by a federal appeals court, and it has been sitting in public since January 2025.

The good news is that the rule is narrower than the headlines suggest, and the fix is a process change rather than a legal one. But you have to know where the line is before you can stay on the right side of it.

What the Copyright Office Actually Said

In January 2025 the US Copyright Office published Copyright and Artificial Intelligence, Part 2: Copyrightability. It is the second in a series, it followed a notice of inquiry that drew more than 10,000 comments from all 50 states and 67 countries, and roughly half of those comments addressed exactly this question.

The report reaches eight conclusions. Four of them matter to anyone commissioning work:

• Using AI tools to assist rather than stand in for human creativity does not affect the copyright protection available for the output.
• Copyright protects the original human expression in a work even where the work also contains AI generated material.
• Copyright does not extend to purely AI generated material, or to material where there is insufficient human control over the expressive elements.
• On current generally available technology, prompts alone do not provide sufficient control.

That fourth point is the one that surprises people, because it holds no matter how much effort went into the prompt. The Office addresses the effort argument directly and rejects it, pointing to Feist Publications v. Rural Telephone Service, the 1991 Supreme Court decision establishing that copyright protects original authorship rather than hard work. The report puts it plainly: revising a prompt repeatedly is re rolling the dice, generating more outputs to choose from without changing your degree of control over any of them.

The Office also ran its own test. It fed a detailed prompt into a commercial image generator asking for a bespectacled cat in a robe reading a newspaper, specifying fog, lighting, composition and lens. The output delivered some of those instructions and ignored others, and added an incongruous human hand holding the newspaper that nothing in the prompt had asked for. Where the instructions ran out, the system filled the gaps itself. That gap between what you specified and what came back is the whole argument.

This is not just an agency view. In Thaler v. Perlmutter, decided 18 March 2025, the DC Circuit held that the Copyright Act requires all eligible work to be authored in the first instance by a human being. The Supreme Court declined to hear an appeal in March 2026, docket number 25-449, which leaves that ruling standing.

One thing Thaler didn’t decide, though: Dr Thaler told the Office his image was created autonomously with no human creative contribution at all, so the court answered only the narrow question of whether a machine can be an author. The harder question, how much human involvement is enough, is still working its way through the courts in a separate case, Allen v. Perlmutter.

Where the Line Sits: Three Things That Do Count

Read only the headline and you would think AI assisted work is uncopyrightable. That is not what the report says. It identifies three specific human contributions that do produce protectable expression.

1. Your own work, perceptible in the output

If you feed in something you created, an illustration, a photograph, a piece of writing, and that work is still visible in what comes out, you own that part. The Office describes the scope as analogous to a derivative work: your expression is protected, the AI generated elements around it are not.

2. Creative selection, coordination and arrangement

Assemble AI generated material into a whole with enough creative judgement and the compilation is protectable, even though the components individually are not. One practical limit from the Office’s own Compendium of Copyright Office Practices: a selection of only two or three elements is generally treated as too minimal to qualify.

3. Creative modification

Edit, adapt or enhance an output enough and your changes carry their own protection. The report specifically notes that tools letting users select and regenerate individual regions of an image, rather than reprompting the whole thing, can amount to genuine control over placement of creative elements in a way that plain prompting does not.

Sitting underneath all three is the assistive use carve out, which is broader than most people realise. The Office lists colour correction, detail sharpening, de-blurring, song ideation and generating a preliminary outline for a piece of writing as uses that do not affect copyright in the finished work at all. If you brainstorm with a model and then write the thing yourself, nothing about your copyright changes.

Two Registrations That Show the Line in Practice

Abstract tests are hard to apply. Two actual Copyright Office decisions make the line concrete, and they went opposite ways for a reason worth understanding.

Zarya of the Dawn. A comic book with Midjourney generated images and human written text. In its February 2023 decision, the Office protected the text and the selection, coordination and arrangement of words and images across the pages. It refused protection for the images themselves. The book is copyrighted. The pictures in it are not.

SURYAST. An applicant uploaded their own photograph and instructed a system to render it in the style of Van Gogh’s The Starry Night. The Office’s Review Board refused registration in December 2023, finding insufficient creative control. Compare that with a work called Rose Enigma, where an artist’s hand drawn illustration remained clearly perceptible in the output. That one was registered, with the registration annotated to cover only the human authorship visible in the deposit.

The difference is not effort or intent. It is whether a human authored work survives visibly into the output. In one case the input got absorbed. In the other it did not.

One more thing that catches people out. The Office’s March 2023 registration guidance requires applicants to disclose AI generated content that is more than de minimis, and to describe the human contribution. This is a disclosure obligation on you, not a question the examiner will guess at. The Office has registered hundreds of works containing AI generated material on exactly that basis.

The UK Rule Is Different, and It May Not Survive

If you sell into the UK, the position is not the same, and this is where a lot of generic advice written for a US audience stops being accurate.

Section 9(3) of the Copyright, Designs and Patents Act 1988 covers literary, dramatic, musical and artistic works generated by computer in circumstances where there is no human author. It assigns authorship to the person by whom the arrangements necessary for the creation of the work are undertaken. Protection runs for 50 years from creation under section 12(7).

That provision predates generative AI by roughly three decades, and no reported UK case has applied it to an AI output. Ireland, New Zealand, Hong Kong and India carry comparable provisions, with the same absence of case law testing them.

It may not last. The UK government ran a copyright and AI consultation from December 2024 to February 2025 that drew over 11,500 responses. Under sections 135 to 137 of the Data (Use and Access) Act 2025, it was required to report to Parliament by 18 March 2026. That report landed on the deadline, and as analysis from Bird & Bird sets out, removing section 9(3) protection was one of the few firm outcomes in it. No legislation has been introduced, and the report is not binding.

The practical read: if you are relying on section 9(3) to claim ownership of AI output in the UK, you are relying on an untested provision that the government has signalled it intends to repeal. That is a thin foundation for an asset you are counting on.

What Changes on 2 August 2026 If You Sell Into Europe

Ownership is one question. Disclosure is a separate one, and the deadline is close.

Article 50 of the EU AI Act, Regulation (EU) 2024/1689, applies from 2 August 2026. It imposes four transparency duties, and it is not limited to high risk systems. Providers of systems that interact directly with people must make clear a person is dealing with AI. Providers of generative AI systems must mark synthetic outputs in a machine readable format, detectable as artificially generated. Deployers must label deepfakes and AI generated text published on matters of public interest.

Two details matter commercially. First, the Article 50(4) labelling duty for AI generated text does not apply where the content has undergone human review or editorial control and a natural or legal person holds editorial responsibility for publication. If you have a real editorial process, you are covered by the terms of the exemption itself. Second, the European Commission published the first draft of a Code of Practice on marking and labelling in December 2025, and the marking obligation sits with the AI provider rather than with you as a user.

Penalties under the Act reach 15 million euro or 3 percent of worldwide annual turnover, whichever is higher.

The Framework: Documenting Authorship Before You Need It

Nobody asks you to prove authorship on a good day. They ask during an infringement dispute, a registration application, or an acquisition where someone is valuing your content library. All three arrive after the work is done, which is exactly when reconstructing what happened is hardest.

Four steps, none of which require a lawyer.

1. Sort output into three buckets, not two

Why it matters: the useful question is not whether AI was used. It is which of the three protected contributions is present.

How it is done: label each asset as assistive use, human input perceptible in output, or selection and arrangement plus modification. Anything that fits none of those is unprotected, and you should know which assets those are before you build a strategy around them.

2. Keep the human authored input, not just the output

Why it matters: Rose Enigma succeeded because the hand drawn illustration was still perceptible in the final image, and the applicant could show it. SURYAST failed because the input did not survive visibly.

How it is done: archive the source illustration, draft, photograph or recording alongside the finished asset. Storage is cheap. Reconstructing an input you deleted eight months ago is not possible.

3. Record the editorial decisions, not the prompts

Why it matters: the Office has been explicit that prompt volume and prompt effort do not establish authorship. What does establish it is selection, arrangement and modification, and those are decisions, not inputs.

How it is done: a short note per asset covering what was chosen from what, how it was arranged, and what was changed afterwards. Two or three lines. It is the same instinct as a version history, applied to editorial judgement rather than code.

4. Fix the disclosure question at brief stage

Why it matters: US registration requires you to disclose more than de minimis AI content, and EU rules require labelling in defined circumstances from August 2026. Both are far easier to answer while the work is being made.

How it is done: add one line to your creative brief template asking whether AI generated material will appear in the deliverable and who holds editorial responsibility for it. That single line answers both regimes.

None of this is a reason to stop using these tools. The Copyright Office went out of its way to say that assistive use changes nothing, and it explicitly declined to recommend new legislation, concluding that existing law handles the question. The exposure is not in using AI. It is in assuming you own something you have never checked.

Pick one asset you shipped last quarter. Ask which of the three buckets it falls into, and whether you could evidence that today. If the answer is no, you have found the gap, and it takes about ten minutes per asset to close it going forward.

FAQ