Copyright in AI-generated content: what applies to businesses?
The legal area is not yet fully settled, but the main picture is reasonably stable: purely AI-generated content (text or images created by a model without significant human creative effort) typically has weak or unclear copyright protection, because copyright as a starting point presupposes a human, original creation. On the other hand, responsibility for how you use the content lies with the business. This article is an honest overview, not legal advice or any kind of guarantee.
Table of contents
A brief look at the starting point
When you let an AI model write a product description or generate an image, three questions arise that managers often mix up:
- Can I use it commercially? (typically yes, but it depends on the supplier's terms)
- Do I own it; can I stop others from copying it? (often unclear for what is purely machine-generated)
- Can I be held liable? (yes, responsibility for the use lies with you)
The three are not the same. You may perfectly well be allowed to use a piece of content commercially without having a strong copyright in it, and without being protected if the output happens to resemble someone else's protected work.
Can purely AI-generated content be protected by copyright?
The short, honest answer: unclear, and for what is purely machine-generated probably no, or only to a limited extent.
Copyright is historically built on the idea of a work that expresses a person's own intellectual creation. When a result has been produced by a model from a short prompt without significant human creative contribution, it is doubtful whether a protected copyright arises at all, and who would be the rights holder if it did.
In practice that means: an image or a text that you generate with a few clicks can in principle be copied by a competitor, without you necessarily having a strong legal basis to take action against them. For marketing material that needs to be your own and to be defended, that is a real consideration.
Be aware at the same time that the rules may vary between countries, and that the area is developing in both a Danish and a European context. What applies today may be clarified.
Why "unclear" and not just "no"
Because a lot of content is in reality not purely machine-generated. As soon as a person makes real creative choices (selects, edits, composes, combines), you are approaching something that may have protection. The boundary is fluid, and that is exactly where many businesses find themselves.
Human input matters
This is the most actionable point in the whole article: the more real human creative effort, the stronger your position.
If an employee uses AI as a draft and then rewrites, structures, adds their own points, selects and edits images, and combines elements into something new, then the content moves from "purely machine-generated" towards "a human-made work in which AI was a tool." A tool you create with is not the same as a machine that creates for you.
There is no fixed percentage threshold for when the input is "enough", and beware of anyone who claims an exact figure. But the direction is clear: superficial use gives a weak position, substantial human effort gives a stronger one.
The risk of output that resembles protected material
Here the risk turns the other way. A model is trained on large amounts of existing material, and in rare cases it can produce output that lies close to an existing, protected work: a passage of text, an illustration style, a recognisable element.
If you use such output commercially, you can in principle come into conflict with a rights holder, even though you didn't knowingly copy anything. For you as a business it makes no difference that "it was the AI"; it is you who published it and profited from it.
Particularly worth watching for:
- Recognisable characters, logos and trademarks. An AI image containing something that resembles a protected brand or a well-known character is a clear risk.
- Faces and people. Images that resemble real, identifiable people also raise questions other than copyright (for example the right to one's own image).
- Text that is strikingly close to a well-known source. If a paragraph sounds as if it was taken directly from an article or book, check it.
Check the supplier's terms: who has the rights?
This is something you can clarify today without a lawyer: read the terms of the AI service you use.
The suppliers (image generators, text models, video tools) typically state in their terms of use what you may do with the output. Look for:
- Commercial use: May the output be used commercially at all? Some free or hobby plans don't allow it.
- Rights to the output: Does the supplier transfer the rights to you, or do they retain a right themselves? Do other users get access to similar output?
- Plan level: The rights often follow the subscription type. A paid business plan often gives broader rights than a free account.
- Disclaimer of liability: Almost all suppliers disclaim liability for whether the output infringes third parties' rights. That responsibility lands with you.
The point: even when the supplier "gives you the rights to the output," it doesn't guarantee that the output is protected by copyright, and it doesn't guarantee that it doesn't infringe on others. They are two different things.
Practical advice for the business
None of these requires a lawyer to get started. They reduce risk and markedly strengthen your position.
1. Edit: use AI as a draft, not as the final answer
Have a person add real creative effort to everything that goes out publicly or is used commercially. That strengthens both the quality and your legal position.
2. Check output before publishing
Introduce a simple check: Does the image resemble a well-known brand, a character or a real person? Does the text sound strikingly close to an existing source? One person's eyes on the material before publication catch most obvious problems.
3. Document the process
Keep the prompts, drafts and the human changes. If doubt later arises about who created what, being able to show that a person made real creative choices is worth its weight in gold.
4. Read and keep the supplier terms
Note which plan you're on and what it allows. Switch to a business plan if the free one doesn't allow commercial use.
5. Draw up a simple internal guideline
One page saying: which tools are approved, what has to be edited, what must not be published directly, and who checks. That makes the risk manageable across the team.
When should you talk to a lawyer or an IP specialist?
The self-help above goes a long way, but seek professional advice when the stakes get high or the matter becomes specific:
- When the content is business-critical: a logo, a campaign or a product that you need to be able to defend as your own against copying.
- In contracts and customer deliverables, if you deliver AI-created material to customers and promise them rights that you're not sure you can give.
- In a specific dispute, if someone claims that your material infringes their rights, or you want to pursue a copier.
- For larger investments, before you build a brand or a product line on top of AI-generated material.
A specialist can assess your specific situation, which a general article never can.
Important reservation
This article is a general overview and not legal advice. The legal area around copyright and AI is developing, practice is not yet fully settled, and the rules may vary between countries and change over time. Nothing here is a guarantee of a particular legal outcome. If you're in doubt about a specific matter, seek professional legal advice from a lawyer or an IP specialist.
How to use AI responsibly in practice
The technical and the legal go hand in hand: a well-thought-out way of using AI (with editing, checking and documentation built in) is both better content and lower risk. That's where we help. Agent360 builds real AI solutions and advises businesses on introducing them responsibly in day-to-day work.
Book a no-obligation AI conversation with Agent360, and we'll look at how your business can use AI for content and operations in a way that creates value without picking up unnecessary risk along the way.
Contents
What takes up your time?
Write down the task that takes up the most time. We reply whether it is worth building, and we say so if it is not.
Write what takes up your timeRelated articles

MFA: multi-factor authentication for your business – how to get started
Learn how to implement MFA securely in your organisation. Guidance on methods, costs and best practices for protecting company data.

English-language invoices: requirements for invoices to foreign customers
Invoice in English to foreign customers legally. Danish regulations, EU and international VAT rules, zero-rating, reverse charge, and automation.

Drawing programme (CAD): 2D, 3D, licences and file formats
What is CAD? Learn about 2D and 3D design programmes, file formats, licensing models, and how to choose the right tool for technical drawings.