Reflections from Alicante on opt-out, the AI Act and the future of protected-content marketplaces.
At a time when generative artificial intelligence is moving faster than regulation, Europe has decided to look the future of copyright straight in the eye. The EUIPO Conference on Copyright 2025, held at the EUIPO headquarters in Alicante, brought together regulators, academics, digital platforms, rightsholders and startups to address one key question:
How can we ensure that AI keeps innovating without leaving creators behind?
From BRANDLEX, we had the opportunity to closely follow both Day 1 and Day 2 of the conference, identifying trends that are reshaping the copyright framework in Europe and that will directly impact tech companies, rightsholders and AI developers.
Copyright as infrastructure for the creative economy
In the opening sessions, one fundamental idea was stressed: copyright is not just a legal rule, but an economic and cultural infrastructure. Copyright-based industries account for around 7% of the EU’s GDP and more than 70 million jobs, and they are undergoing a profound transformation driven by digital technologies and AI.
From the European Commission and the European Parliament, speakers underlined the need to adapt the legal framework so that:
- Creators continue to receive fair remuneration.
- Tech companies and AI developers have clear rules to innovate.
- Citizens can access culture, education and research in a safe and balanced environment.
Opt-out, TDM and the new balance between AI and creators
One of the core topics on Day 2 was the discussion around “opt-in / opt-out” models and text and data mining (TDM):
- The European model is currently based on opt-out: by default, works can be used for TDM unless the rightsholder has explicitly reserved their rights under the Copyright in the Digital Single Market Directive (CDSM/DSM).
- This opt-out must be expressed clearly and, preferably, in a machine-readable manner, through legal notices, metadata or specific technical solutions.
- Where an opt-out is in place, any use for training AI systems requires an authorization or licence.
A major challenge identified at the conference is the lack of common standards to record and effectively enforce these opt-outs. Today, fragmented legal and technical solutions coexist, making compliance difficult for AI providers and limiting control for rightsholders.
The AI Act: transparency, traceability and respect for opt-outs
The other major piece of the puzzle is the European AI Act, which introduces specific obligations for providers of general-purpose AI models:
- Respect the opt-outs declared by rightsholders.
- Publish sufficiently detailed summaries of the data used to train their models.
- Move towards greater traceability of the protected content used, so that training datasets and processes can be audited.
In practice, this means that AI training is no longer a “black box”:
organisations that develop or integrate AI will need to know what data they use, under which conditions and on what legal or licensing basis.
Licensing and marketplaces: towards a new market for protected data and content
A particularly relevant discussion for BRANDLEX focused on licensing and the emergence of future content marketplaces:
- Exercising the opt-out does not block the use of works altogether, but it forces the negotiation of licences when they are to be used for AI training.
- This is driving the rise of digital solutions and specialised marketplaces that connect rightsholders (publishers, record labels, producers, database owners, etc.) with AI developers who need high-quality datasets.
- For this market to function, opt-out mechanisms must be simple, robust and accessible, allowing both large and small rightsholders to participate, declare reservations and negotiate terms.
The conference also explored complementary models such as:
- Extended collective licensing managed by collective management organisations.
- Statutory remuneration schemes or “AI levies”, which would allow creators to be compensated when their works are used on a large scale in AI systems.
Many of these models are still under discussion, but the direction is clear: AI training creates value, and that value should be shared more fairly across all agents in the creative and technological ecosystem.
Creators, diversity and education: the human dimension of copyright
The conference was not just a technical debate. Creators, artists, writers, musicians and designers shared their experiences regarding:
- The precariousness that still affects many authors.
- The use of their works, images or voices in AI systems without clear information or consent.
- Concerns that new technological tools could widen the gap between large platforms and small creators.
Speakers insisted on the importance of:
- Investing in copyright education from schools through to universities.
- Protecting Europe’s linguistic and cultural diversity in the face of AI models trained predominantly in just a few languages.
- Involving creators in the design of public policies and licensing models.
The Copyright Knowledge Centre: a new European hub
One of the most relevant announcements was the launch of the EUIPO Copyright Knowledge Centre, conceived as a space for:
- Research and production of studies.
- Training and practical materials.
- Best practices and coordination among Member States, institutions, rightsholders, platforms and civil society.
This Centre is set to become a European reference point for navigating copyright challenges in the age of AI and the digital economy.
What does all this mean for companies – and for BRANDLEX?
For European and Latin American companies working with data, content or AI, the conference’s main takeaways can be summarised in three ideas:
- AI leaves a legal footprint: training models will no longer be a grey area. Traceability, clear agreements and robust licensing strategies will be essential.
- Opt-out is only the beginning: real systems will be needed to manage reservations of rights, licences and remuneration.
- Marketplaces and specialised solutions will be key: regulatory complexity opens the door for platforms that make compliance and licence negotiation more transparent and secure.
At BRANDLEX, as a firm specialised in intellectual property, technology law and generative AI, we see this evolution as a major opportunity to:
- Support companies and rightsholders in designing copyright & AI strategies.
- Develop LegalTech solutions that connect licensing, regulatory compliance and data-driven business models.
- Build bridges between Europe and Latin America, facilitating regulatory softlanding for companies that want to operate on both sides of the Atlantic.
Shall we talk?
If your company is:
- Training AI models with protected data or content,
- Designing products that integrate generative AI, or
- Exploring new licensing models for intangible assets,
BRANDLEX can help you align innovation, copyright and regulatory compliance.
👉 Write to us at info@brand-lex.com or visit our website to learn how we can work together.