In the US, copyright protects the parts of your AI-assisted marketing that a person authored: your writing, your selection and arrangement, and your edits. A prompt on its own does not count. In the UK, the law has a special rule for computer-generated work, and the government proposed in March 2026 to remove it. In both countries, keep a dated record of what your people did.
That does not make AI-drafted content unusable. It changes what you can stop a competitor from copying, and the answer depends on how much of the work a person did.
How many small businesses are putting AI into their marketing?
Enough that the ownership question is practical, not academic. A research brief from the Federal Reserve Bank of San Francisco, dated July 2026, found that about 40% of respondents to the 2024 Small Business Credit Survey reported either using or planning to use AI. The survey covers US firms with fewer than 500 employees. A total of 13,608 firms responded between September and November 2024, and more than 8,500 answered the AI question. The brief lists marketing, social media and search engine optimization among the common applications, but it gives no percentage for marketing alone, and "using or planning to use" is not the same as "using today".
One more detail from the brief matters here: respondents noted that marketing and SEO tools are often built into the platforms sellers already use. That means a product description or a social post can be AI-assisted without anyone deciding it should be. If that sounds familiar, shadow AI in a small business is the same pattern in a different place.
What does US law say about AI-generated content?
US law protects human authorship, and the Copyright Office has said what that means for AI output. The courts have been clear on the first half. In Thaler v. Perlmutter, the D.C. Circuit held on 18 March 2025 that "the Copyright Act of 1976 requires all eligible work to be authored in the first instance by a human being." The Supreme Court declined to hear the appeal on 2 March 2026 (docket 25-449), so that ruling stands. The case involved a work that listed a machine as its only author, which is not how a business uses AI, but it fixes the starting point.
The more useful document for an owner is the US Copyright Office's Copyright and Artificial Intelligence, Part 2: Copyrightability, published on 29 January 2025. Three of its points apply to marketing teams:
- On prompts: "Based on the functioning of current generally available technology, prompts do not alone provide sufficient control." Typing a good instruction is not authorship.
- On what you add: human authors are entitled to copyright in their works of authorship that are perceptible in AI-generated outputs. Your own paragraph, photo or illustration, sitting inside the output, keeps its protection.
- On how you assemble it: protection can reach the creative selection, coordination or arrangement of material in the outputs, or creative modifications of the outputs.
There is a practical consequence if you ever register a work. The Office's registration guidance, published on 16 March 2023, says applicants "have a duty to disclose the inclusion of AI-generated content in a work submitted for registration," and that AI-generated content that is more than de minimis "should be explicitly excluded from the application." What you can claim is what a person did, so the record of what a person did is the asset.
What does UK law say?
UK law has a rule that US law does not, and the government wants to remove it. Section 9(3) of the Copyright, Designs and Patents Act 1988 says that for a computer-generated literary, dramatic, musical or artistic work, "the author shall be taken to be the person by whom the arrangements necessary for the creation of the work are undertaken." Section 178 defines "computer-generated" as work generated by computer "in circumstances such that there is no human author of the work," and section 12(7) sets the term at 50 years from the end of the calendar year in which the work was made. (All three as shown on legislation.gov.uk on 2 November 2026.)
In its Report on Copyright and Artificial Intelligence, published on 18 March 2026, the government says that this protection "departs from the core rationale for copyright, which is to encourage and reward human creativity." Its proposal, at paragraph 47: "in the absence of evidence of its ongoing value, we propose that this specific type of protection should be removed, while copyright should continue to protect works created with AI assistance."
That is a proposal, not law. Nothing in the parts of the report I read draws the line between a work with "no human author" and an "AI-assisted" one for an ordinary business prompt, so I would not guess where your content falls. On labelling, the report says that "there are no obligations in the UK for AI-generated content to be labelled" and proposes to work with industry on best practice.
How do the two approaches compare?
The two countries start from different rules and end up pointing the same way: the part you can most safely call yours is the part a person did. Here is the comparison, checked at the sources on 2 November 2026.
| Question | United States | United Kingdom |
|---|---|---|
| Who can be an author? | A human being (D.C. Circuit, 18 March 2025; Supreme Court review denied 2 March 2026) | A human, plus a special rule for computer-generated work: the person who made the arrangements (section 9(3)) |
| Output with no human authorship | Not protected. Prompts alone do not give sufficient control (Copyright Office, 29 January 2025) | Covered today by the computer-generated works rule, for 50 years (section 12(7)); government proposed removal in March 2026 |
| AI-assisted work with real human input | Protected for the human-authored parts, and for creative selection, arrangement and modification | Government says copyright should continue to protect works created with AI assistance |
| Disclosure | Registration applicants must disclose and exclude more than de minimis AI content | No obligation to label AI-generated content; best practice to be explored with industry |
| Direction of travel | Settled for now: human authorship is required | Proposed change, not yet law |
Verdict: if your content leans on a person's choices and edits, you stand on firmer ground in both countries. If it is raw output, the US gives you little to enforce and the UK gives you a rule that may not last. I have compared two jurisdictions and one question, who holds rights in what you publish. Whether an AI tool's output copies someone else's work is a separate risk that I have not covered here. This is a guide to the sources, not legal advice.
What I Tell Business Owners
Across the business teams I train, the first question about AI-made content is almost always "can we use it?" The second question comes later, or never: "could someone copy this and would we be able to stop them?" I tell owners to ask the second question early for the handful of pieces that carry the brand, and to stop worrying about it for everything else.
The law's answer rhymes with a principle I use everywhere: outsource the doing, not the thinking. In both countries, the thinking is the part that carries weight: the decisions, the structure, the examples only you have.
Which of your content is worth protecting? The Fingerprint Test
Most of what a business publishes does not need protecting, and a few pieces do. The Fingerprint Test is my way of sorting them. It is my suggested way of thinking, not established research.
The Fingerprint Test is three questions to ask of any piece of marketing content before you rely on owning it: Can you point to what a person decided, chose or wrote in it? Can you show it, with a dated record? And would it matter if a competitor copied it word for word? Yes to the first two means you hold something you can defend. No to the third means you can stop worrying and get on.
Here is how the test sorts a typical small business's content.
| Content | Does a person's choice show in it? | What to keep | Worth protecting? |
|---|---|---|---|
| Signature guide, white paper or course notes | It should: the structure, examples and opinions are yours | Outline, AI draft, edited version, dates, who edited | Yes |
| Core website pages and sales emails | Usually, if someone rewrote the draft | Final copy and the version before it | Yes, for the pages that win business |
| Brand story and founder pieces | Yes, because it comes from you | Your notes or recording that fed the draft | Yes |
| Routine social posts and captions | Rarely | Nothing special | No |
| Bulk product descriptions | Rarely | Nothing special | Seldom |
| Ad images and video | Only if a person chose, arranged or reworked the result | The brief, the selection and the edits | Check the tool's terms first |
Names, logos and brand marks sit under different law (trademark and design). Ask an intellectual property adviser about those; this article is only about the copyright in the content itself.
What if an agency or freelancer makes your content?
Ask them the five questions in the Name-On-It Rule, which I set out in what to ask suppliers who use AI:
The Name-On-It Rule is one rule for any work you buy that goes out under your name or touches your customers' information: the supplier may use AI on it, but only after they have told you in writing where they use it, what of yours goes into it, who checks the output, who answers if it is wrong, and what happens to your material when the job ends. If your name is on the work, you need those five answers before the work starts, because your customer will ask you, not your supplier.
For content, add one more request: a short note of which parts a person on their team wrote or changed, with dates. A contract that assigns rights to you can only pass on whatever rights exist in the first place, so the supplier's record of human contribution is what makes the assignment worth having. Ask your solicitor to look at the wording.
What should you do this week?
You can do all of this in an afternoon, without a lawyer or a project.
- List the ten pieces of content that carry your brand: your flagship pages, guides, signature offers and best-performing emails.
- Run the Fingerprint Test on each, and mark the ones where a person's choices show.
- For those, start keeping the trail in one shared folder: the brief, the AI draft, the edited version, who edited it and the date. A dated file name is enough.
- Read the terms of the AI tools and platforms your team uses for what they say about ownership of outputs, and write down what you find.
- Put the question into your supplier briefs, and into your AI policy if you have one (do you need an AI policy? and the UK small business AI policy template cover where it goes).
- Teach the people who edit to edit for substance. The edit is what creates the human contribution, and training staff to check AI output is the same skill.
- Diary a review in six months. The UK proposal is not law yet, and this area is moving.
Telling customers that content was made with AI is a separate question from who owns it; AI disclosure for a small business covers that. For the wider picture of marketing work, see AI for sales and marketing in a small team.
Where to go next
If your team is producing content with AI and nobody has agreed who edits what, that is exactly what my AI training for teams is built for; organizations I have worked with include Google, Grammarly and Hyve Group. If you want to know where your business stands first, take the Business AI Readiness Scorecard, and if you are planning an event, my business keynotes are here.
Sources and further reading
- Federal Reserve Bank of San Francisco, AI Adoption Among Small Businesses: Qualitative Insights from the Small Business Credit Survey, Community Development Research Brief 2026-01, July 2026 (survey of US firms with fewer than 500 employees, fielded September to November 2024).
- US Copyright Office, Copyright and Artificial Intelligence, Part 2: Copyrightability, 29 January 2025.
- US Copyright Office, registration guidance for works containing AI-generated material (policy statement dated 10 March 2023, published 16 March 2023, 88 Fed. Reg. 16,190).
- US Court of Appeals for the D.C. Circuit, Thaler v. Perlmutter, No. 23-5233, decided 18 March 2025.
- Supreme Court of the United States, Thaler v. Perlmutter, No. 25-449 docket, petition denied 2 March 2026.
- UK legislation, Copyright, Designs and Patents Act 1988, section 9, section 12 and section 178, as shown on 2 November 2026.
- UK Government, Report on Copyright and Artificial Intelligence, GOV.UK, 18 March 2026.
Dan Fitzpatrick is The AI Educator: a keynote speaker, Forbes contributor and bestselling author who has trained more than 150,000 people across 30+ countries.
