AI & Digital Assets
We advise on the intellectual property issues raised by artificial intelligence and digital assets: who owns what is generated with AI, what data a model may be trained on, and how rights are structured when an asset is represented, sold or licensed in digital or tokenised form. This is the area of our practice where European regulation, particularly the EU AI Act, meets traditional intellectual property doctrine most directly. We do not build technology: we design the legal structure of the rights that technology represents or transfers.
Scope of work
Rights in AI output
Contractual allocation of rights, authorship risk, chain of title and exploitation rights for AI-generated or AI-assisted output, between developers, users and commissioning parties.
Training data rights
Legal review of datasets used to train AI models, with particular attention to copyright, database rights and the text and data mining provisions of the Directive on Copyright in the Digital Single Market (Directive (EU) 2019/790).
AI use policies
Drafting and review of internal AI use policies: how rights in generated output are handled, what information may be shared with a model and what must be reviewed before use.
Asset tokenisation
Legal structuring of tokenisation: which rights the token represents, who holds them, what licence the buyer receives and how those rights are exploited and enforced. The technology itself is delivered by the client’s specialist providers.
Authenticity, provenance & chain of title
Legal analysis and documentation of the provenance and chain of title of works and digital assets, and of any certificates or authentication systems the client uses, so that what they certify has a sound legal basis.
Physical products with a digital identity
The legal link between physical products and their digital rights or versions (certificates, digital twins, collectibles) in fashion, luxury, art and consumer goods, in coordination with our trade mark and design practices.
Digital licensing & exploitation
Licences and royalties for digital content and assets, including the terms attached to tokens and smart contracts, reviewed from an intellectual property standpoint.
Coordination with META Channel Corporation
Coordination with META Channel Corporation’s regulatory team where a matter also involves the AI Act, MiCA, the GDPR or DORA.
This is the area of intellectual property where the law is moving fastest. We move with it.
When to engage us
Integrating generative AI
A company integrating generative AI into its products or operations and needing to allocate rights over generated output by contract.
Tokenising an asset
A business preparing to tokenise an asset or sell it in digital form, and needing to define which rights it transfers.
Works in training data
A rights holder concerned about the use of its works in AI training datasets.
Internal AI policy
A corporate client requiring an internal policy on AI use that allocates intellectual property risk correctly.
Physical product with a digital version
A fashion, luxury or consumer brand linking its products to digital certificates or digital versions.
Jurisdictions
European Union (AI Act, Directive (EU) 2019/790, GDPR), Spanish national law, and coordinated work in Latin American and African jurisdictions where digital asset regulation is emerging.
Frequently asked questions
Who owns content created with generative AI?
There is no automatic answer, which is precisely the risk. The tool’s terms do not settle the question, and the law may not grant copyright at all where human input is limited. What a contract can do is allocate, between the developer, the user and any commissioning party, whatever rights and risks exist. We address this through clear contractual allocation, chain-of-title and exploitation-rights clauses, rather than assuming ownership exists by default.
Can AI-generated content be protected by copyright?
In the EU, copyright protects creations that reflect a human author’s own intellectual choices, so output produced with little or no human creative input carries a real authorship risk and may not be protectable as such. The practical answer is to document the human contribution and to secure the value contractually, so that what you rely on does not depend solely on whether the raw output qualifies for copyright.
What is the EU AI Act and does it affect my intellectual property?
The EU AI Act is the Union’s regulation of artificial intelligence, applying on a phased timeline. It is primarily a product-safety and governance framework rather than an IP statute, but it intersects with intellectual property in areas such as transparency about AI-generated content and the data used to train models. We advise on how those obligations affect your IP position, coordinating with META Channel Corporation’s regulatory team.
Can I use copyrighted data to train an AI model?
In the EU, the text and data mining provisions of Directive (EU) 2019/790 allow mining in defined circumstances, but rights holders can reserve their works from such mining through an express opt-out (Article 4(3) of the Directive), and database rights may also apply. Whether a given dataset is safe to use depends on its sources and on any reservations attached to them. We review training datasets and the rights position before the model is trained.
What does tokenising an asset mean in legal terms?
Tokenising means representing an asset or a right through a transferable digital record. Legally, what matters is not the technology but what the token represents: ownership of an object, a licence to use it, a share of revenue or simply a certificate. If that is not properly defined, the buyer may believe it is acquiring rights it does not acquire. We structure the transaction so that the token, the contract and the intellectual property rights all say the same thing, and coordinate the regulatory analysis where the asset may fall within MiCA.
If I buy an NFT or another tokenised asset, do I own the copyright in the work?
Generally not. Buying the token transfers the token, not the copyright in the work it refers to; the buyer acquires the copyright only if it is expressly assigned in writing; otherwise it may use the work only within whatever licence the terms grant. The rights a buyer actually receives depend entirely on the terms attached to the token, which is why we structure those terms so that they say what the parties intend.
Can a digital certificate prove that a product or a work is authentic?
It can help, but it only proves what the documentation behind it supports: who the author or manufacturer is, who has owned it and which rights have been transferred. We therefore review the chain of title and the documents underpinning the certificate, and how they fit with the client’s trade marks, designs and copyright. We do not issue certificates or run the technology: we make sure they rest on a sound legal basis.
How should I allocate rights over AI output in my contracts?
By addressing it expressly, rather than leaving it to a default that may not exist. The contract should allocate rights and risk between developer, user and commissioning party, establish the chain of title, and define exploitation rights, warranties and indemnities around the output. We draft these clauses for AI-assisted work so each party knows what it can use, license and defend.
Does my business need an internal AI use policy?
For most organisations using AI tools, yes. A policy sets out who holds rights over generated output, what may be disclosed to a model, and what must be flagged or reviewed before use, which both protects confidential information and reduces intellectual property risk. We review and draft internal AI use policies and align them with your contracts and regulatory obligations.
How does this connect with MiCA, GDPR and DORA?
AI and digital-asset matters rarely sit within intellectual property alone: some tokenised assets may fall within MiCA, training data and AI systems may fall within the GDPR where personal data is involved, and the operational resilience of financial entities is subject to DORA. Because IP Global Guard is part of META Channel Corporation, we coordinate these overlapping questions within the same organisation, so the IP position and the wider regulatory position stay consistent.