Verified as of 27 September 2026. If anything has changed, let me know and I will update it with the date of the change.
More and more artists use artificial intelligence tools at some stage of their work: for sketching, generating a starting image, varying a composition, or producing the entire piece. The practical question is always the same: is the result protected by copyright and, if so, to whom does it belong?
This guide explains how Spanish law answers it as of 27 September 2026. It is not legal advice. Some points, especially the threshold of human intervention required, remain open.
The starting point: the author is a person
Article 5 of the Intellectual Property Act defines the author as ‘the natural person who creates any literary, artistic or scientific work’. Article 10 protects original creations expressed by any means or medium. Taken together, these provisions mean that copyright protection requires an original human creation.
An AI tool cannot be an author or the original holder of copyright. This makes it possible to distinguish two situations.
A result produced without sufficient human creative contribution is not protected as a work merely by virtue of having been generated. Writing a prompt and accepting the first output may not suffice to attribute the creation of the result to a person. In that case, whoever used the tool does not automatically obtain copyright over the image simply for having requested it. This does not rule out restrictions arising from a contract, from rights in materials incorporated into the result, or from rules other than copyright.
Human creative contributions can be protected. If a person selects, arranges, combines, or transforms materials through free and creative decisions that are reflected in the result, those contributions may be protectable. Depending on the case, protection may apply to the final work, to its human elements, or to the original selection and arrangement, not necessarily to each piece of generated material it contains.
Where the boundary lies
Writing a few words and accepting the first result is not the same as developing a process in which a person tests variants, selects, corrects, combines materials, and shapes a piece of their own. Between these two extremes lies a wide area, and that is where much real-world work sits.
I have not found a Spanish court ruling that sets the threshold of human intervention for a work created with generative AI. Nor can it be stated, as a general rule, that an elaborate prompt is sufficient on its own. The decisive issue will be whether the result expresses identifiable human creative decisions, rather than merely the effort invested or the technical complexity of the process.
Selecting among results deserves the same care. Choosing one image from several does not automatically protect the chosen image. The selection or arrangement may be original, but it will be necessary to assess which decisions the person made and how they are reflected in the final result.
What to do in practice
- Document the process. Keep prompts, intermediate versions, discarded outputs, layers, editing files and dates. This documentation can help show which decisions you made and which parts come from your own intervention.
- Identify what is generated and what is transformed. Distinguishing the source material from your modifications lets you pin down which contributions you could defend as your own.
- Do not confuse contract with copyright. The terms of a tool may grant or restrict uses between the company and the user, but they do not in themselves create copyright where there is no protectable human creation.
- Check the provenance of the result. Having no rights of your own in an image does not make it legally harmless. The result could reproduce protected elements, trademarks, designs, or identifiable features belonging to third parties.
- Disclose it in sales and commissions when relevant. Explaining that AI was used, what part of the process it played, and what rights are being assigned reduces the risk of misunderstandings regarding authorship and the scope of the licence.
- Agree terms precisely. In a commission, it is advisable to keep separate the physical delivery, the permissions to use the work and, where they exist, the intellectual property rights being transferred.
What this question does not settle
That an image generated without sufficient human contribution is not protected by copyright means that no one obtains that exclusive right over it merely by generating it. It does not mean that it can be used without regard for contracts or third-party rights. Nor does it resolve whether the image is art.
The law needs to identify a protectable creation and the person to whom it is to be attributed. The question of what makes a work artistic belongs to a different level. I developed it in ‘AI art requires no author’ and in my research on structural surplus.
On the open conversation
This guide summarises the treatment of authorship of works made with artificial intelligence tools in Spanish legislation and the available legal scholarship. Where sufficient human contribution begins is still an open question. If you work in copyright or use these tools in your practice and wish to contribute a case or a nuance, the guide remains open.
Sources
- Royal Legislative Decree 1/1996, of 12 April, approving the consolidated text of the Intellectual Property Act, arts. 1, 5, 10 and 12 (in Spanish).
- Cuatrecasas, ‘Inteligencia artificial y derechos de autor’ (in Spanish).
- DiG Abogados, ‘Derechos de autor en obras generadas por inteligencia artificial: ¿quién es el dueño de la obra?’ (in Spanish).
- LegalToday, ‘Crear con inteligencia artificial también tiene dueño’, 22 September 2026 (in Spanish).