A technology company based in California deploys a generative AI system to produce marketing copy, product imagery, and software modules for its Mexican subsidiary. The output is substantial, commercially valuable, and produced without a single human hand drafting a line. When the company moves to register those assets in Mexico and license them to a regional distributor, it discovers that the country's intellectual property legislation was written for a world in which authors are human. The question of who owns the work – and whether it is protected at all – has no clear statutory answer.
AI-generated works in Mexico occupy a legal grey zone under current intellectual property legislation, which requires a human author for copyright to vest. The Instituto Nacional del Derecho de Autor (National Copyright Institute of Mexico, known as INDAUTOR) has not issued binding guidance specifically addressing AI authorship. International businesses deploying AI tools in Mexico must therefore rely on contractual structuring, technology licensing agreements, and documented human creative oversight to preserve enforceable rights.
This analysis examines the doctrinal foundations of Mexico's copyright regime, the gap between statute and current practice, competing interpretations emerging across the Americas. The cross-border dimension for international operators. Additionally, the strategic steps available now. before a definitive legislative response arrives.
Doctrinal foundations: what Mexican intellectual property law currently says
Mexico's intellectual property legislation – rooted in the federal law governing copyright and related rights – defines a work as a creation of the human intellect expressed in a specific form. This definition is not incidental. It reflects a broader civil law tradition in which authorship is an attribute of personhood. Moral rights – the right to claim authorship, to protect the integrity of the work, and to oppose modifications – are inalienable and can only be held by a natural person.
The statute identifies several categories of protected works: literary, musical, dramatic, choreographic, pictorial, sculptural, cinematographic, photographic, architectural, and software, among others. Crucially, it also recognises works produced by an employer or on commission, where the economic rights vest in the commissioning party while moral rights remain with the human creator. This employer-work doctrine is the closest existing instrument to address AI-assisted creation.
Software receives specific protection under Mexican intellectual property legislation as a literary work. The code, architecture, and documentation generated by an AI system could, in principle, fall within this category – provided a human author can be identified. Where no such identification is possible, the statutory basis for protection collapses.
The Suprema Corte de Justicia de la Nación (Supreme Court of Mexico) has not yet issued a definitive ruling on AI authorship. Lower courts and administrative bodies have dealt with tangential questions. such as the protectability of algorithmic outputs in the context of database rights and software liability. but none has addressed the core authorship question head-on. This absence of authoritative jurisprudence creates both risk and opportunity for international practitioners.
Practitioners in Mexico note that INDAUTOR's registration examiners currently apply a default requirement: a natural person must be named as author. Applicants seeking to register AI-generated works have encountered informal rejections or requests for supplementary documentation demonstrating human creative contribution. This administrative practice, while not codified, shapes the real-world availability of copyright protection for AI output in Mexico today.
The human-authorship gap: where statute and practice diverge
The gap between the letter of Mexican intellectual property law and the reality of AI-assisted production is widening rapidly. Generative AI systems now produce outputs that range from lightly curated human-directed creations to fully autonomous generations with minimal human input. Mexican copyright doctrine was not designed to distinguish between these scenarios.
Three positions have emerged in academic commentary and early administrative practice. The first holds that AI output with no human creative contribution is entirely unprotectable – it is neither authored nor ownerless in any legally meaningful sense, and falls outside the copyright system entirely. This position has the strongest textual basis in current legislation but produces commercially uncomfortable results: a competitor can freely copy the output.
The second position argues that the human who directs, trains, curates, or selects AI output exercises sufficient creative judgment to qualify as an author of the resulting work. Under this view, the degree of human intellectual contribution is the operative variable. A product designer who selects among fifty AI-generated images, makes modifications, and integrates the result into a commercial product may have a viable authorship claim over the final selection and arrangement.
The third position, advanced by some technology licensing specialists. Proposes that the commissioning entity. typically the company that deploys the AI system. should hold economic rights by analogy to the employer-work doctrine, even without a named human author. This argument has not been tested before Mexican courts. However, it finds indirect support in the legislature's own treatment of collective works and anonymous works. Both of which vest rights in entities rather than named individuals.
De jure, the human-author requirement governs. De facto, administrative practice at INDAUTOR has shown willingness to engage with documentation that demonstrates meaningful human oversight in the production process. A company that can present a detailed record of human creative decisions – prompts, iterations, selection criteria, editorial revisions – is better positioned than one that submits AI output without any such record.
For businesses operating across the Americas, this doctrinal uncertainty has a concrete cost. A regional brand deploying AI-generated product imagery cannot be certain that its Mexican registrations will withstand a cancellation challenge. A software developer whose AI system writes functional code for a Mexican client faces questions about whether that code is protectable, who owns it, and what algorithmic accountability obligations arise if the code causes harm. These are not hypothetical risks. They are immediate questions with commercial stakes.
For a tailored strategy on technology licensing and AI-related intellectual property protection in Mexico, reach out to info@ferrazwhitmore.com.
Competing interpretations across the Americas and the EU regulatory signal
Mexico does not make intellectual property policy in isolation. Its copyright doctrine sits within a network of international obligations – the Berne Convention, the TRIPS Agreement, and the bilateral trade commitments arising from its major free trade arrangements. None of these international instruments expressly addresses AI authorship, but all influence how Mexico's courts and legislature are likely to respond as pressure builds.
The United States has taken an early position through its copyright administrative practice. The relevant authority has declined to register works produced autonomously by AI systems, while accepting applications for works where human creative selection and arrangement are documented. This approach – protecting the human layer, not the AI output – is directionally consistent with where Mexican doctrine appears to be heading. A comparative reading of AI-generated works and intellectual property in the United States reveals how this threshold is being calibrated in a closely watched common law jurisdiction. Additionally. It provides a useful reference point for Mexico-based practitioners advising US-linked clients.
Brazil has begun legislative consultations on AI regulation that include explicit provisions on authorship of AI-generated output. The dominant Brazilian proposal assigns rights to the deployer of the AI system, subject to disclosure obligations. If enacted, this approach would create a divergence between Brazilian and Mexican law that cross-border operators would need to manage explicitly in their technology licensing and distribution agreements.
The European Union's AI Act – which has attracted significant compliance interest among multinationals – introduces a risk-based regulatory system for AI applications, including obligations around transparency, documentation, and algorithmic accountability. While the AI Act does not directly govern intellectual property rights, its disclosure and record-keeping requirements create a documentary infrastructure that could support human-authorship claims in civil law jurisdictions outside the EU. A company subject to AI Act compliance obligations in Europe will, by necessity, produce the kind of human-oversight documentation that Mexican copyright practice currently demands informally.
This regulatory convergence matters strategically. Companies building AI Act compliance programmes for their European operations should extend those documentation standards to their Mexican deployments. The incremental cost is low. The benefit – a defensible record of human creative oversight for copyright purposes in Mexico – is significant.
Software liability is a related dimension. Under Mexican civil and commercial legislation, a party that deploys a software system causing damage to a third party may face liability regardless of whether the software was autonomously generated. The absence of copyright protection for AI-generated code does not insulate the deployer from tortious or contractual liability for that code's behaviour. Practitioners in Mexico note that the software liability analysis applies independently of the authorship question – a point that many technology clients initially overlook.
Digital services operating in Mexico that rely on AI-generated content face a further layer of exposure. If the output of an AI system reproduces elements of a protected third-party work. a real risk in systems trained on large datasets. the deployer may face infringement claims even if the AI-generated output itself lacks copyright protection. The absence of protection for the AI output does not create a safe harbour against claims arising from the training data. Mexico's intellectual property legislation on reproduction rights and neighbouring rights applies to this scenario, and INDAUTOR has competence to receive complaints in this area.
To discuss how AI-generated content risks and technology licensing strategies apply to your cross-border operations in Mexico, contact us at info@ferrazwhitmore.com.
Strategic recommendations for international operators in Mexico
International businesses deploying AI tools in Mexico should not wait for legislative reform. The commercial risk of inaction is concrete: without proactive structuring, AI-generated assets may be unregistrable, unenforceable against copyists, and vulnerable to third-party claims. The following strategic framework addresses the gap between current doctrine and commercial necessity.
Document human creative contribution at every stage. The most defensible copyright position under current Mexican law rests on demonstrating that a human author exercised genuine creative judgment. This means retaining records of prompts, iterative selections, editorial decisions, and integration choices. Where multiple team members contribute to the direction and curation of AI output, a collective authorship structure may be available under Mexican intellectual property legislation.
Use technology licensing agreements to allocate AI-output rights contractually. Where statutory copyright protection is uncertain, contractual rights fill the gap between counterparties. A well-drafted technology licensing agreement can specify that the commissioning party holds exclusive rights to use, reproduce, and commercialise AI-generated output within defined territories and for defined purposes. These rights are enforceable between contracting parties even where they cannot be asserted against third parties under copyright law.
Align deployment practices with AI Act compliance standards. As noted above, the documentation required for AI Act compliance in Europe. transparency records. Human oversight logs, risk assessments. maps closely onto the evidentiary requirements that Mexican administrative practice informally applies to AI-generated copyright applications. Building a single compliance infrastructure that satisfies both sets of demands is achievable and cost-efficient for companies with operations in both regions.
Register what is registrable now. Where a human author can be identified – even partially – registration with INDAUTOR should be pursued. Registration in Mexico is declaratory, not constitutive, but it creates a strong evidentiary presumption that courts and enforcement authorities will respect. Early registration also establishes priority against subsequent competing claims.
Address training data provenance before deployment. AI systems trained on third-party content carry embedded infringement risk. Before deploying an AI system in Mexico, operators should conduct a provenance assessment of the training dataset. Where material was ingested without a licence, a remediation strategy – re-training on licensed or public domain material, or obtaining retroactive licences – should be implemented before commercialisation.
For companies working across the Americas, Mexico's position sits at a pivotal intersection. The intellectual property practice in Mexico at Ferraz & Whitmore provides integrated advice across copyright registration, technology licensing, and enforcement strategy for AI-related matters.
Outlook: what to monitor and how the legislative trajectory is likely to develop
Mexico's legislature has begun to receive proposals addressing AI and intellectual property. The political dynamics are familiar: industry stakeholders push for a deployer-rights model that protects AI-generated output. creative communities resist any departure from human authorship requirements. and government agencies seek alignment with international treaty obligations that were drafted before generative AI existed.
The most likely near-term development is not a comprehensive AI authorship statute but a targeted amendment to existing intellectual property legislation. That amendment is expected to address three discrete questions: the protectability of AI-assisted works where human contribution is documented. The allocation of rights between AI developers and deployers. Additionally, the disclosure obligations for AI-generated content in commercial contexts. Legislative consultations have engaged INDAUTOR, the federal competition authority, and the telecommunications regulator – a sign that the reform will cut across several branches of Mexican technology and commercial legislation.
Administrative guidance from INDAUTOR is likely to arrive before legislative reform. Practitioners in Mexico anticipate that INDAUTOR will issue a circular or interpretive guideline establishing evidentiary standards for AI-related copyright applications. This administrative instrument, while not binding on courts, would significantly reduce the uncertainty that currently confronts applicants at the registration stage.
Courts will eventually be called upon to address the authorship question directly. The route to the Supreme Court of Mexico in intellectual property matters runs through the federal courts system via amparo proceedings. a constitutional review mechanism available to any party whose rights under federal law are at stake. A well-structured test case, filed by a party with a documented and commercially significant AI-generated work, could produce a judicial interpretation of the human-authorship requirement within two to three years. Several technology companies operating in Mexico are understood to be monitoring this route.
The algorithmic accountability dimension will also develop independently of the authorship question. As AI systems are deployed in financial services, healthcare, and public administration in Mexico, the absence of a dedicated digital services liability regime is becoming visible. Existing civil and commercial legislation can accommodate many of these claims, but the procedural and evidentiary tools available to plaintiffs are imperfect fits. Reform in this area is likely to precede copyright reform, given its broader political salience.
For international businesses, the practical message is this: the window between now and Mexico's legislative response is not a period of legal safety. It is a period of maximum exposure to rights uncertainty. Companies that build defensible documentation practices, structured technology licensing agreements, and AI Act-aligned oversight processes now will be positioned to convert those investments into registered rights as soon as the legislative regime clarifies. Companies that defer action risk discovering, at the moment they most need enforcement, that their AI-generated assets carry no enforceable protection.
Frequently asked questions
Q: Can an AI system be listed as the author of a work registered in Mexico?
A: No. Mexican intellectual property legislation requires a human author for any work to receive copyright protection. An AI system cannot hold authorship or exercise moral rights. Registration in Mexico must name a natural person or legal entity as the rights holder, typically the developer or the commissioning party.
Q: How long does it take to register an AI-assisted work with the Mexican copyright authority?
A: Registration with the Instituto Nacional del Derecho de Autor (INDAUTOR) typically takes between four and twelve weeks, depending on document completeness and the category of work. Expedited procedures are available in limited circumstances. Registration is declaratory in nature, meaning rights arise at creation, but registration creates a strong evidentiary presumption of ownership.
Q: Is it a misconception that AI output is automatically in the public domain in Mexico?
A: It is a common misconception that AI-generated output falls automatically into the public domain in Mexico. The legal position is unsettled. Where a human author exercised sufficient creative input in directing or selecting AI output, copyright protection may attach to that human contribution. Works with no identifiable human creative input occupy a legal grey zone that Mexican courts have not yet definitively resolved.
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 technology licensing across the Americas and beyond. We work with international entrepreneurs, institutional investors, and in-house legal teams who need results-oriented counsel on AI-generated content, algorithmic accountability, software liability, and digital services regulation. As an international law firm in Mexico and across Latin America, Ferraz & Whitmore supports clients navigating the gap between current statute and fast-moving AI practice. Our attorneys have advised on technology licensing and IP enforcement matters across both civil law and common law systems. The firm's practice covers 15 practice areas across 46 jurisdictions, with direct experience before Mexican administrative bodies and across Iberian and Americas markets. Engaging a lawyer in Mexico with cross-border AI experience at the right stage protects both your registered rights and your downstream licensing position. To discuss your AI-generated works situation in Mexico, 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.