There is no single worldwide answer to who owns AI-generated work. Copyright rules differ between countries, and the outcome can depend on how much creative work a person contributed, what the AI produced and what the relevant contract says. After Part 2’s discussion of the creative spark, this instalment turns to what follows creativity: authorship, permission and credit.
First, “ownership” can mean several things
When someone asks who owns an AI-generated image, article or track, they may be asking different questions.
- Copyright asks whether the work qualifies for legal protection, and who holds the rights.
- Permission asks whether someone is allowed to use the work, which may depend on licences, contracts and the AI service’s terms.
- Credit asks whose contribution should be recognised.
- Value asks who benefits when creative work is made, used, sold or used to develop AI.
These questions overlap, but they are not interchangeable. A service’s terms may give a user permission to use an output without guaranteeing that the user owns copyright in it. And a person can deserve credit for a contribution even when the legal status of the finished work is uncertain.
What copyright currently says in the UK, US and EU
The UK has a distinctive rule for computer-generated works
In the UK, the Copyright, Designs and Patents Act 1988 includes a provision for certain works generated by a computer “in circumstances such that there is no human author”. Under section 9(3), the person who undertook the necessary arrangements for creating the work is treated as its author. The Act currently provides a 50-year term for this category.
That is a distinct route, not the same as ordinary copyright in a human-created work using AI as a tool. The Government’s March 2026 Report on Copyright and Artificial Intelligence says that the Government proposes removing this specific protection for wholly computer-generated works, while continuing to protect AI-assisted works with human creativity. A proposal is not a change in the law by itself. The report describes policy direction; section 9(3) should not be treated as repealed unless legislation brings that change into effect.
For work where a person makes meaningful creative choices, for example, a photographer composing and editing an image with AI assistance: the human-authored contribution may qualify for ordinary copyright protection.
The US focuses on human authorship
The US Copyright Office’s report on copyrightability says that copyright protects human-authored expression, including creative human selection, arrangement or modification of AI-generated material. It concludes that, with current generally available technology, prompts alone do not usually give a user sufficient control over the expressive details of an output.
As the Office puts it, “prompts alone do not provide sufficient control”. The key question is not simply whether someone spent time prompting, but what expressive elements they actually created or controlled.
The EU AI Act is not an authorship test
EU copyright protection also centres on human creative contribution. Separately, the EU AI Act requires providers of general-purpose AI models to have a policy for complying with EU copyright law and to publish a sufficiently detailed summary of training content.
Those transparency duties do not decide who owns a particular generated image or grant copyright to its user. They concern model providers and training information, rather than settling authorship for each output.
The work behind the output matters too
Generative systems learn from large collections of material, including writing, photographs, illustrations and music made by people. The UK Government’s report explains that training datasets often contain copyright works gathered from a range of sources. The legal questions around how particular material was accessed and used are still developing and vary between jurisdictions.
This makes the issue bigger than whether an output looks original. Human work helped make these systems useful in the first place, while the resulting tools may produce material that competes with the work of human creators. Writers, designers, photographers and musicians are therefore asking not only whether their work was copied, but whether its use was authorised, acknowledged or fairly rewarded.
The legal test is specific. Copyright generally protects original expression, not an idea, subject or broad style on its own. But if an output reproduces a substantial part of a protected work, that may raise a separate copyright question. A user’s right to use an output under a platform’s terms does not automatically resolve that question.
A design commission needs more than a final file
For a business commissioning a logo, illustration, brochure or campaign, the practical question is often, “Can we use this, and what rights are we receiving?”
A finished PDF or image file does not answer that on its own. If the work is made by a freelance designer, paying for the commission does not necessarily transfer copyright. The parties should agree in writing whether the business receives a licence or an assignment of rights, and what that arrangement covers.
It is worth being specific about:
- Where the design can be used, such as print, a website, social media or advertising
- Whether the permission is exclusive, and for how long
- Whether the business may adapt the design or commission further changes
- Whether editable source files are included
- Which fonts, stock images, illustrations or other third-party assets have separate licence terms
- Whether AI-generated elements are included, and how those elements may be used
These details matter when a business refreshes its branding, reprints a brochure or hands artwork to another supplier. Good records make future decisions clearer for both the commissioning organisation and the creator.

Credit and value are not the same as legal ownership
A designer may set the brief, develop the layout, choose colours, edit an image, reject unsuitable results and prepare the final artwork for print. Those choices can carry substantial professional value, even if some elements were generated by AI.
Likewise, a writer who uses an AI tool for research or drafting may still contribute the structure, reporting, judgement, wording and final edits that shape a publishable article. In music, human composition, performance, production and arrangement may sit alongside generated elements.
The final output can hide the work that made it useful. If credit goes only to the tool, or if a business assumes that “AI-generated” means “free of rights”, human contributions and third-party permissions can disappear from view. At the same time, a clear record of who did what helps organisations talk fairly about authorship and avoid promising rights they do not hold.
Keep a record of the creative process
Provenance is the trail of information showing where an asset came from and how it was developed. It does not prove every legal claim by itself, but it can make a project’s history easier to explain.
For commissioned design, writing, photography or artwork, it is useful to keep:
- The original brief and agreed scope of work
- Early sketches, drafts, layouts and significant revisions
- The final approved files and the date of approval
- A note of any AI tools used and what they contributed
- Licences or permissions for fonts, stock images and supplied material
- The contract or written record setting out usage rights and ownership
- Any required attribution or credit information
For a printed project, preserve the approved artwork and its specifications alongside the relevant permissions. That can help when the same design is adapted for a new format or reprinted later. Our graphic design work and in-house print production bring those stages into the same conversation, but rights and usage should still be agreed clearly for each project.

A practical checklist before using AI-assisted work
- Check the contract: confirm whether you have a licence or an assignment, and what it permits.
- Review the tool’s terms: look for conditions on commercial use, attribution and permitted outputs.
- Keep evidence of human input: save drafts, edits, selections and the final approved version.
- Check third-party material: confirm permissions for stock images, fonts, music or client-supplied assets.
- Look for recognisable copied content: an output that reproduces existing protected expression may need further review.
- Be accurate about credit: describe the human and AI contributions honestly, especially where attribution matters.
This is practical guidance, not legal advice. Where a project has significant commercial or legal risk, an intellectual-property professional can advise on the specific facts and agreements.
The takeaway
The answer to “who owns what AI creates?” depends on the work, the human contribution, the jurisdiction and the contract. In the UK, a specific rule for computer-generated works remains part of the law unless and until it is changed; in the US and EU, human authorship is central to copyright in creative expression.
For businesses, clear agreements and good provenance records are practical ways to understand what can be used, what can be owned and whose contribution deserves recognition. The deeper question is whether the value created with human work is being shared and credited as carefully as the output is being produced.
When a creative work is made with AI, what would fair credit and fair value look like to you?
