A technology company launches a marketing campaign across Spain using images, copy, and music produced entirely by generative AI tools. The campaign is a commercial success. Six months later, a competitor reproduces the same assets. The company reaches for intellectual property protection – and discovers that Spanish law may not recognise ownership of those works at all. The opportunity captured through innovation could be lost through a doctrinal gap that the legislature has yet to close.
AI-generated works occupy an uncertain legal position under Spanish intellectual property legislation. Spanish law conditions copyright protection on a human author, meaning works produced autonomously by an AI system may not qualify for traditional copyright. Businesses operating in Spain that rely on AI-generated content must understand this gap now, before competitors or courts define the rules against them.
This analysis examines the doctrinal foundations of Spanish copyright law as applied to AI-generated output, reviews the competing interpretations emerging from legal practice. Identifies the gap between formal legislation and commercial reality, maps the cross-border dimension for European clients. Additionally, offers a strategic outlook for businesses navigating this evolving area.
The doctrinal foundations: authorship, creativity, and the human requirement
Spanish intellectual property legislation rests on a foundational principle: copyright belongs to the human being who creates an original work. This is not a peripheral rule. It is the structural core of the entire protective system. The author is defined by personal, intellectual contribution. Without a human being who makes creative choices, the legislation's protective apparatus does not engage.
This doctrinal position was inherited from continental civil law traditions that predate digital technology by several centuries. The French and German influences on Spanish intellectual property law both treated authorship as an expression of personality. A work was protectable because it reflected the mind of its creator. That philosophical foundation creates immediate tension with AI-generated output, where the generative act is performed by a machine operating on statistical patterns rather than personal expression.
Under Spanish intellectual property legislation, the Tribunal Supremo (Supreme Court of Spain) has consistently held that originality requires a human creative decision. Works are protectable when they express the author's individual stamp. Where that stamp is absent – because no human made the relevant choices about form, arrangement, or expression – the work falls outside the protective scope.
For AI-generated works, this creates a layered problem. The human beings involved in producing an AI-generated output include: the developers who built the AI system, the trainers who shaped its behaviour, and the end-users who issued the prompt. None of them, under current doctrine, is straightforwardly the author of the output. The developer created the tool, not the work. The trainer shaped the system's tendencies. The prompt-writer specified an outcome but did not make the detailed expressive choices that constitute authorship in the doctrinal sense.
Practitioners in Spain note that this creates a doctrinal vacuum rather than a clear negative rule. The legislation does not explicitly exclude AI-generated works. It simply fails to accommodate them. That gap is commercially significant. A work that is not protected is, in principle, freely available to any third party who wishes to use it.
Competing interpretations and the gap between statute and practice
Spanish courts have not yet produced a definitive ruling on AI authorship. The absence of binding authority has allowed several competing interpretive approaches to develop in legal practice. Understanding these approaches is essential for any business that creates, licenses, or relies on AI-generated content in the Spanish market.
The first approach – call it the tool theory – treats the AI system as an instrument in the hands of a human author. Under this view, the prompt-writer or the user who directs the AI is analogous to a photographer who uses a camera. The camera does not author the photograph; the photographer does, through the choices of framing, timing, and subject. Applied to AI, this theory would attribute authorship to the human who shaped the generative process through prompting, parameter setting, and selection among outputs. Spanish legal commentators who favour this approach argue that it is consistent with existing legislation and avoids the destabilising conclusion that vast categories of commercial content are unprotected.
The second approach – the autonomy threshold theory – draws a distinction based on the degree of human involvement. Where a human makes meaningful creative choices at multiple stages of the generative process, the result may be protectable. Where the AI operates with near-complete autonomy on a minimal prompt, the human contribution is too thin to ground authorship. This approach has practical appeal because it maps onto observable differences in how AI tools are actually used. A designer who iterates through dozens of versions, selects, adjusts, and combines outputs is doing something qualitatively different from a user who types four words and accepts the first result.
The third approach – the public domain by default theory – holds that autonomous AI output is unprotectable and that the appropriate response is acceptance rather than doctrinal contortion. Under this view, the correct position is that AI-generated works belong to the public domain, available for all. Proponents argue that this outcome is commercially disruptive in the short term but beneficial in the long term: it lowers barriers to creative reuse. Prevents monopolisation of AI-generated content by large technology companies. Additionally, reflects an honest reading of legislation that was not designed for this situation.
In practice, businesses face a situation where no single theory commands judicial authority. The Tribunal Supremo has addressed adjacent questions about computer-generated works in earlier periods, but these decisions predate modern generative AI and their application to current systems is genuinely contested. Lower courts in Spain – including the Audiencia Provincial (Provincial Court) level – have produced divergent outcomes in matters touching on digital and software-generated content.
The gap between formal legislation and commercial practice is already wide. Many Spanish businesses are treating AI-generated content as proprietary, asserting ownership through contractual mechanisms even where copyright protection is legally uncertain. Licensing agreements, employment contracts, and platform terms of service regularly purport to assign rights in AI-generated works. These arrangements may provide practical protection in a world where most disputes do not reach litigation. But they are vulnerable the moment a well-resourced party decides to challenge them.
For a tailored strategy on protecting AI-generated content in Spain, reach out to our AI and technology law practice in Spain.
Adjacent protections: software, databases, and unfair competition
Because primary copyright protection is uncertain, practitioners in Spain have developed a set of adjacent legal strategies for protecting AI-generated works. These alternatives do not fully replicate copyright protection, but they provide meaningful commercial value in many scenarios.
Software copyright is the most direct alternative. The AI system itself – the underlying code, architecture, and trained model – qualifies for protection under software copyright provisions within Spanish intellectual property legislation. This protects the tool rather than its output, but it is legally sound and enforceable. A competitor who copies the AI system, or who attempts to replicate its outputs by reverse-engineering the model, may infringe software copyright even if the individual outputs are unprotected.
Technology licensing plays a central role here. Businesses can structure their commercial relationships so that the AI system itself is the licensed asset. Clients and partners receive a right to use outputs generated by a protected system, rather than receiving ownership of the outputs themselves. This structure sidesteps the authorship question while maintaining commercial control. It also aligns with the algorithmic accountability principles emerging from EU AI regulation, which focus responsibility on the deployer and provider of the AI system.
Database rights provide another avenue. Where an AI system generates a large collection of content – product descriptions, training data annotations, classification results – that collection may qualify as a database under Spanish and EU law. Database protection does not require originality in individual entries. It protects the investment made in obtaining, verifying, or presenting the collection's contents. For businesses that generate AI content at scale, this can be a practically significant protection.
Unfair competition legislation in Spain offers a further protective layer. Even where a work is unprotected by copyright, systematic copying of a competitor's AI-generated content may constitute unfair competition if it creates confusion in the market, exploits another's investment, or amounts to slavish imitation. The threshold for unfair competition claims is different from copyright – it does not require a finding of authorship – and it has been applied by Spanish courts in a range of digital content contexts.
The combination of software copyright, technology licensing, database rights, and unfair competition claims creates a defence-in-depth strategy for businesses that cannot rely on conventional copyright. None of these mechanisms is as strong as a clear copyright claim. But together, they provide a commercially meaningful protective position while the primary doctrinal questions remain unresolved.
Digital services companies operating in Spain should also consider trade secret protection. Where the AI system's architecture, training data, or operational parameters give competitive advantage, those elements may qualify for protection under Spanish trade secret legislation. This protection is available regardless of the copyright status of the outputs. It requires active steps to maintain secrecy, but it can be an effective tool against competitors who attempt to replicate the AI system's performance.
The EU AI Act and its interaction with Spanish intellectual property law
The EU AI Act has introduced a new layer of regulatory requirements that intersect with intellectual property questions in ways that Spanish businesses need to understand. The Act does not directly resolve the authorship question. It is a safety and transparency regulation, not a copyright instrument. But its requirements have important downstream effects on how AI-generated works are produced, documented, and deployed in the Spanish market.
AI Act compliance requirements apply to providers and deployers of AI systems operating in the EU. For high-risk AI systems, the Act mandates technical documentation, conformity assessments, and human oversight measures. For general-purpose AI models – the large language models and image generators at the heart of most AI-generated content – the Act imposes transparency obligations and. There. Those models are trained on copyrighted material, copyright compliance requirements including disclosure of training data.
This last point has significant practical implications. If an AI system was trained on copyrighted works without adequate licences, any output produced by that system may carry legal risk beyond the authorship question. The original rights holders in the training data may have claims against the system provider. Those claims are being actively litigated across multiple European jurisdictions. Spanish courts will inevitably be drawn into this body of litigation as cases involving Spanish-language content and Spanish market deployment accumulate.
The interaction between AI Act compliance and intellectual property strategy is already shaping commercial practice. Businesses that procure AI-generated content are beginning to require contractual warranties from AI system providers – representations that the system was trained on properly licensed data and that outputs do not infringe third-party rights. These warranty structures shift risk back toward providers. They are becoming standard in enterprise-level software liability arrangements in Spain and across the EU.
For a comparative view of how Portugal has addressed similar questions under its intellectual property legislation, our analysis of AI-generated works and IP in Portugal covers the parallel doctrinal developments under the Portuguese system.
The EU AI Act also introduces transparency requirements for AI-generated content specifically. Content produced by AI systems – particularly where it could be mistaken for human-created content – must in certain contexts be labelled as AI-generated. This labelling obligation is separate from copyright status. A work can be AI-generated in a legally disclosed and compliant way and still be commercially valuable. But the obligation does affect how businesses present their AI-generated assets to consumers and business partners in Spain.
Cross-border implications for European clients
Spain is not alone in facing these questions, but the Spanish legal system's specific contours produce outcomes that differ meaningfully from those in other European jurisdictions. Businesses operating across multiple EU markets cannot assume that their IP strategy for AI-generated works will transfer automatically from one jurisdiction to another.
The UK's approach is instructive. English copyright legislation contains an explicit provision for computer-generated works. It attributes authorship to the person who makes the arrangements necessary for the creation of a work. This provision, developed before generative AI existed, is now being interpreted in the context of modern AI systems. Its existence gives UK-based businesses a statutory foothold that their Spanish counterparts lack. A business that generates AI content for both markets faces the paradox that the same output may be protected in the UK and unprotected in Spain. or vice versa. Depending on how courts develop the doctrine.
Germany, France, and the Netherlands – the other major EU economies that generate substantial AI-related litigation – have each taken somewhat different approaches to the human authorship requirement. German intellectual property legislation, like Spain's, is grounded in a strong personality rights tradition that is resistant to non-human authorship. French law has shown some willingness to recognise protection for computer-generated works through judicial creativity, though no definitive legislative solution exists. Dutch practice has been pragmatic, with commercial practice running ahead of formal doctrine.
For a business structuring its AI content strategy across the EU, the practical implication is that ownership must be documented at the point of creation, not assumed. Contracts with AI system providers, employment agreements with human operators, and internal governance procedures should all address IP ownership explicitly. The corporate vehicle used to hold IP rights also matters. A Sociedad Anónima (SA – Spanish public limited company) or Sociedad de Responsabilidad Limitada (SL – Spanish private limited company) holding AI-generated assets as trade secrets or database rights may be in a stronger position than one relying on undefined copyright claims. Proper corporate structuring, documented through a Notario (notary public) and registered in the Registro Mercantil (Spanish commercial registry), supports the enforceability of any IP assignments or licensing arrangements.
The enforcement environment also varies across borders. Spain's court system – including the specialised IP courts and ultimately the Tribunal Supremo – has shown increasing sophistication in technology-related disputes. But litigation timelines in Spain remain longer than in some Northern European jurisdictions. For many businesses, the practical enforcement question is whether a copyright claim will produce a remedy before the commercial value of the AI-generated content has dissipated. Speed of protection matters as much as the theoretical scope of rights.
To discuss how IP and AI regulation interact for your business across European jurisdictions, contact us at our intellectual property practice in Spain.
Strategic outlook and recommendations
The doctrinal uncertainty around AI-generated works in Spain is unlikely to resolve quickly through judicial decision. The Tribunal Supremo tends to address issues once a sufficient body of lower-court decisions has accumulated. At the lower court level, cases raising the AI authorship question directly are still relatively rare. Legislative reform at the EU level – through a possible revision of the Copyright Directive or a separate AI-specific instrument – is being discussed but will take years to produce binding rules.
In the interim, businesses face a strategic choice: treat AI-generated content as unprotected and price that risk into their commercial decisions, or pursue an active IP strategy using the adjacent mechanisms described above. The right answer depends on the commercial value of the content, the competitive landscape, and the risk appetite of the business.
Several strategic principles emerge from current practice and the doctrinal analysis above.
Document human creative input at every stage. The tool theory of AI authorship – which attributes protection to the human who directs the AI – requires evidence of genuine human creative decisions. Businesses should maintain records of prompting iterations, selection choices, and post-generation editing. These records support a copyright claim and, if copyright fails, establish the basis for a trade secret or unfair competition argument.
Structure contracts to allocate risk explicitly. Agreements with AI system providers should include warranties on training data provenance and output indemnification. Employment contracts and contractor agreements should address ownership of AI-generated works clearly, specifying who holds rights in AI-assisted output. Licensing arrangements should define the scope of permitted use without relying on unstated assumptions about copyright status.
Pursue AI Act compliance proactively. The AI Act's transparency and documentation requirements, though distinct from copyright, are increasingly referenced in commercial negotiations and disputes. A business that can demonstrate rigorous AI Act compliance – including training data disclosure, human oversight procedures, and risk classification – is in a stronger position both commercially and legally. Compliance documentation also supports the argument that human oversight was sufficient to ground authorship under the tool theory.
Monitor the Tribunal Supremo and the EU legislature. The first major Spanish court decision on AI authorship will shift the doctrinal landscape significantly. Businesses with material exposure to this question should be monitoring judicial developments and be prepared to adjust their IP structures quickly. Similarly, any EU legislative initiative on AI and copyright will require rapid implementation across all EU market strategies.
Consider IP holding structures carefully. Where AI-generated content has significant long-term value, the corporate vehicle used to hold it matters. The choice between an SA and an SL involves considerations of capital structure, governance, and transferability of assets. Where cross-border licensing is anticipated, the holding structure may also have tax implications that require analysis alongside the IP strategy. Early structuring decisions, properly executed through a notary and registered with the Registro Mercantil, create a cleaner chain of title and support enforcement in later disputes.
The businesses that position themselves most effectively during this period of legal uncertainty are those that treat the question as a strategic issue rather than a compliance one. The absence of clear rules creates risk, but it also creates opportunity. A business that builds robust documentation practices, thoughtful contractual protections. Additionally. Proactive AI Act compliance now will be better placed when the courts and legislature eventually provide clarity. whether that clarity confirms or denies copyright protection for AI-generated works.
Frequently asked questions
Q: Can a Spanish company own copyright in AI-generated images or text under current law?
A: Under current Spanish intellectual property legislation, copyright requires a human author. AI-generated works produced without meaningful human creative input are unlikely to qualify for traditional copyright protection. However, where a human being directs the AI, selects among outputs, and makes identifiable creative choices, a copyright claim is arguable under the tool theory of authorship. though no definitive court ruling confirms this position. Businesses should supplement any copyright claim with contractual and adjacent protections as a precaution. Engaging a lawyer in Spain with experience in IP and technology law is advisable before treating AI-generated assets as fully protected.
Q: How long does it take to register a database right or software copyright in Spain?
A: Spain does not require formal registration for copyright or database rights to arise – protection is automatic under intellectual property legislation once the qualifying criteria are met. However, voluntary registration with the Spanish copyright registry provides important evidential benefits and simplifies enforcement. The registration process typically takes several weeks. For database rights, the key requirement is documenting the substantial investment made in creating the database. For software, source code and technical documentation should be preserved from the date of creation to establish priority.
Q: Does the EU AI Act affect IP ownership of AI-generated works?
A: The EU AI Act does not directly address copyright ownership of AI-generated works. It is a safety and transparency regulation. However, it requires providers of general-purpose AI models to disclose training data used, which affects claims by original rights holders whose works may have been used without licence. AI Act compliance also creates documentation practices – human oversight records, risk assessments, technical documentation – that can support IP ownership arguments under the tool theory of authorship. Businesses operating in Spain should treat AI Act compliance and IP strategy as complementary, not separate, workstreams.
About Ferraz & Whitmore
Ferraz & Whitmore is an international law firm based in Lisbon, advising business clients across 46 jurisdictions. Our team combines Portuguese civil law expertise with English common law tradition to deliver cross-border legal solutions in AI and technology law, intellectual property protection, and digital services regulation. In Spain, we advise technology companies, institutional investors, and in-house legal teams on AI Act compliance, software liability, technology licensing, and IP strategy across both civil law and common law systems. Our practitioners have advised on AI-related matters before regulatory authorities and in commercial disputes across multiple European jurisdictions. As an international law firm in Spain and across the EU, Ferraz & Whitmore helps clients build enforceable IP positions in a legal environment that is changing faster than the legislation can follow. To discuss how the evolving rules on AI-generated works affect your business, contact us at info@ferrazwhitmore.com.
Disclaimer: This publication is provided for informational purposes only and does not constitute legal advice. The information herein should not be relied upon as a substitute for professional legal counsel tailored to your specific circumstances. Ferraz & Whitmore assumes no liability for actions taken or not taken based on the contents of this material. For advice regarding your particular situation, please contact info@ferrazwhitmore.com.