Fuelling success with legal knowledge.

24 Upper Brook Street London W1K 7QB +44 (0) 333 444 5544 info@maybrooklaw.com LinkedIn Instagram Twitter
Back to top

Who Owns AI-Generated Content? Copyright in AI Outputs in the UK, EU and US

Key takeaways

  • UK: The UK expressly protects “computer-generated” works with no human author in the traditional sense, but how this applies to generative AI is uncertain, and the Government has proposed removing the protection.
  • US: US copyright law requires a human author. Purely AI-generated material is not protected, although human selection, arrangement or modification of AI output can be.
  • EU: In the EU, protection depends on a work being its author’s own intellectual creation. National courts have so far refused protection for images and logos generated from prompts.
  • Contracts and records: AI providers’ terms can allocate whatever rights exist in outputs but cannot create copyright where none exists. Records of human creative input are increasingly important.

Businesses increasingly use AI to generate reports, marketing copy, software code, images, designs and other content. An investment platform may produce AI-drafted research summaries for clients; a product team may use AI to generate code or interface designs; a marketing team may use AI to create campaign images. In each case, a commercially important question arises: does the business own the output, and can it stop others from copying it?

This question is separate from whether AI training or AI outputs infringe third-party copyright, which we consider in our article on AI training and copyright. Here the focus is on whether copyright protects AI-generated content and, if so, who owns it. The answer differs significantly between the UK, EU and US, and in each jurisdiction depends heavily on the level of human creative input involved.

Why does ownership of AI-generated content matter?

If AI-generated content is not protected by copyright, a business may be unable to prevent competitors from copying or reusing it. That can affect the value of content-driven products, the strength of a brand, and the position of a business in commercial negotiations. It also matters in transactions and investment: a business that warrants that it owns the intellectual property in its products, or that its deliverables are original, may need to consider whether those statements remain accurate where substantial content has been generated by AI.

Can AI-generated content be protected by copyright in the UK?

The general rule: originality and human authorship

Copyright protects original literary, dramatic, musical and artistic works. In THJ Systems Ltd v Sheridan [2023] EWCA Civ 1354, the Court of Appeal considered the originality of software-generated displays. It confirmed that the relevant test for originality is whether a work constitutes the author’s own intellectual creation, as established in Infopaq International A/S v Danske Dagblades Forening (Case C-5/08 EU:C:2009:465). A work is original if it is its author’s own intellectual creation, meaning that the author was able to express their creative abilities by making free and creative choices so as to stamp the work with their personal touch. Although the originality threshold is relatively low, the test is framed by an author’s creative contribution.

Where a person uses AI as a tool and makes their own creative choices – for example, by substantially editing, selecting or arranging AI-generated material – the ordinary rules apply and that person may be the author of the resulting work. Where the work is created by an employee in the course of employment, the employer will generally be the first owner of the copyright.

Computer-generated works

The UK is unusual in also expressly protecting “computer-generated” works, defined in section 178 of the Copyright, Designs and Patents Act 1988 as works generated by computer in circumstances such that there is no human author, in the traditional sense. Under section 9(3), the author of a computer-generated literary, dramatic, musical or artistic work is taken to be the person by whom the arrangements necessary for its creation are undertaken. Protection lasts for 50 years from the end of the calendar year in which the work was made, rather than for the author’s life plus 70 years, and the moral rights to be identified as author and to object to derogatory treatment do not apply.

The provision was introduced long before generative AI and has rarely been considered by the courts. In Nova Productions Ltd v Mazooma Games Ltd [2006] EWHC 24 (Ch), the court found that the frames generated by a video game were treated as computer-generated works whose author was the programmer who devised the game, not the player. How the provision applies to modern generative AI is uncertain in two respects.

  • Who makes the arrangements? It may not be clear whether the necessary arrangements are undertaken by the developer of the model, the provider of the service, or the user who enters the prompts.
  • Can the work be original? Section 9(3) applies to computer-generated works, but the originality test is framed by reference to an author’s free and creative choices. It is therefore unclear whether a work with no human creative input can satisfy the requirement of originality at all.

Proposed reform

In its March 2026 Report on Copyright and Artificial Intelligence, published following its 2024–25 consultation, the Government noted the apparent contradiction between section 9(3) and the originality requirement. The Government has proposed that, in the absence of evidence of its ongoing value, this specific protection for wholly computer-generated works should be removed, while copyright should continue to protect works created with AI assistance. The Government intends to continue monitoring the use and impact of the protection, however, to date, no updated legislation has been introduced. Section 9(3) may therefore be an uncertain basis on which to claim ownership of AI-generated content.

Can AI-generated content be protected by copyright in the UK?

Human authorship is required

In the US, copyright requires human authorship. In Thaler v Perlmutter (March 2025), the US Court of Appeals for the D.C. Circuit upheld the Copyright Office’s refusal to register an image listing an AI system as its sole author, holding that the Copyright Act requires a human author. In March 2026, the US Supreme Court declined to hear a further appeal: as the applicant had maintained that no human creative input had been involved, the case did not address the level of human creative input required for an AI-assisted work to qualify for copyright protection.

The Copyright Office approach

In its January 2025 report on copyrightability, the US Copyright Office concluded 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. Copyright may nevertheless protect human-authored expression that is perceptible in an AI-assisted work, the creative selection, coordination or arrangement of AI-generated material, and creative modifications of AI output, assessed case by case.

The Copyright Office’s decisions illustrate where the line currently falls. For the graphic novel Zarya of the Dawn (2023), registration was limited to the human-authored text and the selection and arrangement of the images, but not the individual AI-generated images themselves. In January 2025, the Office registered A Single Piece of American Cheese, an image produced using AI inpainting tools, on the basis of the human selection, coordination and arrangement of AI-generated material, with the AI-generated elements excluded from the claim; the applicant had documented each step of the creative process. By contrast, the Office refused to register Théâtre D’opéra Spatial, created using more than 600 Midjourney prompts, and that refusal is being challenged in Allen v Perlmutter in the District of Colorado, which was still pending at the date of this article.

Applicants registering works that include more than minimal AI-generated material are expected to disclose it and to exclude it from their claim. Under the work-made-for-hire doctrine, where an employee creates a protectable AI-assisted work within the scope of their employment, the employer is treated as the author.

Can AI-generated content be protected by copyright in the EU?

EU law does not contain a provision equivalent to the UK’s computer-generated works regime. Instead, originality is governed by the case law of the Court of Justice of the EU: a work must be its author’s own intellectual creation, reflecting free and creative choices (Infopaq, C-5/08; Painer, C-145/10). This is generally understood to require human input, and the Court of Justice has not yet ruled on the protection of AI-generated outputs.

The Court of Justice is yet to resolve the copyrightability of AI-generated outputs. National courts that have considered the question have refused protection for outputs generated from prompts. In a decision on October 2023, the Municipal Court in Prague held that an image generated by DALL-E from a short prompt was not a work of authorship under Czech law, because only a natural person can be an author, and treated the prompt itself as an unprotectable idea; the claimant had also been unable to prove that the image was generated from his prompt. A first-instance Munich Local Court decision reported in early 2026 similarly refused protection for AI-generated logos, holding that a dominant human creative influence must be demonstrable and that the effort or cost involved does not by itself justify protection. Italy’s 2025 AI law protects works created with AI assistance only where they are the result of the author’s intellectual effort.

As in the UK and US, a person who uses AI as a tool and contributes their own creative choices, for example by substantially adapting AI output, may obtain protection for their own contribution.

What role do AI providers’ terms play?

AI providers’ terms commonly assign to the user, or disclaim, any rights the provider has in outputs. OpenAI’s terms, for example, provide that, as between the user and OpenAI and to the extent permitted by applicable law, the user owns the output, and assign to the user OpenAI’s rights in the output “if any”. The same terms state that output may not be unique and that other users may receive a similar output.

These terms allocate rights between the provider and the user, but they cannot create copyright where none exists, and they do not bind third parties. A business that receives an assignment of AI-generated output may therefore have contractual comfort as against the provider, but no copyright with which to stop a competitor from using the same or similar content. Terms differ between providers and between consumer and enterprise services, including as to IP indemnities and permitted uses, and should be reviewed for each tool used.

Are there any other ways to protect AI-generated content?

Where copyright is unavailable or uncertain, other forms of protection may assist. Trade marks and registered designs depend on distinctiveness or novelty rather than authorship, and may protect AI-generated logos or product designs if the relevant requirements are met. Confidentiality and trade secret protection may protect AI-generated material that is not published, and contractual restrictions can limit how customers and partners use content supplied to them.

Managing ownership of AI-generated content: a practical checklist

For businesses that rely on AI-generated content, managing ownership risk is likely to involve:

  • Identifying valuable outputs. Consider which AI-generated content is commercially important and where exclusivity matters.
  • Human creative input. Where protection is important, structure workflows so that people make and can demonstrate meaningful creative choices in selecting, arranging and editing AI output.
  • Keep records of prompts, iterations, edits and the human contribution to significant works, since the ability to evidence authorship may be decisive.
  • Provider terms. Review the ownership, similarity, licence and indemnity provisions of each AI service used.
  • Employees and contractors. Ensure employment and consultancy agreements address AI-assisted work, including assignment of any rights and disclosure of AI use.
  • Commercial agreements and transactions. Consider whether warranties of ownership or originality in customer, supplier and investment documents need to reflect the use of AI.
  • Alternative protection. Consider trade mark, design, confidentiality and contractual protection where copyright may not be available.

These questions are best revisited as the law develops: the UK Government’s proposals on computer-generated works, the outcome of Allen v Perlmutter and further decisions in the EU may all change the position.

At Maybrook Law, we advise businesses developing, deploying and investing in AI technologies on the legal and commercial issues arising throughout the development and commercialisation of their products, including intellectual property, technology development arrangements and the use of third-party models, software and data. If you are using AI to create content or products, or would like to discuss any of the issues raised in this article, we would be pleased to advise on your position.

This article reflects the law as at 1 October 2026. Our insights, articles and guides do not, and are not intended to, constitute legal advice or be an exhaustive review of all legal developments. Although every effort is made to ensure that the information provided in this article is accurate as of the publication date, please be aware that this area of law is developing rapidly and may be subject to change. Please seek legal advice before applying the information provided to any specific circumstances, transactions or legal issues.