A united warning from Europe’s creative workforce
The European Writers’ Council (EWC) has joined a joint letter signed by 14 federations representing authors and performers, addressed to the European Parliament’s Committee on Culture and Education (CULT). Its central demand is direct: European policy must uphold the ART principle in the AI era.
ART stands for Authorisation, Remuneration and Transparency. Those three terms may sound procedural, but they go to the heart of whether writers, translators, musicians, actors, voice artists and other creators can continue to earn a living from their work as generative AI becomes embedded in publishing, media, advertising, search and entertainment.
This is not simply a dispute about whether AI is innovative. Most creative organisations recognise that technology can assist research, accessibility, translation workflows and production. The issue is the economic model behind many AI systems: creative works may be copied, processed and used to train commercial models, while the people who created those works struggle to discover whether their material was used, to withhold permission, or to receive payment.
For readers interested in a European alternative to extractive platform economics, this intervention matters. It asks whether Europe will build AI markets around consent, traceability and negotiated value—or accept a system in which cultural labour becomes free raw material for a small number of technology companies.
What the ART principle means in practice
Authorisation: creators should have a real choice
Authorisation means that using protected works to train or operate AI systems should not be treated as an automatic entitlement. A novelist’s book, a screenwriter’s script, a performer’s recorded voice or a translator’s work is not merely data waiting to be absorbed into a model.
In practical terms, authorisation requires rights holders to be able to say yes or no before commercial AI use occurs, or to grant permission through workable collective licensing schemes. This is especially important where individual creators lack the bargaining power or technical capacity to negotiate with a global AI developer.
The difficulty is that existing EU copyright rules include text-and-data-mining exceptions. Rights holders may be able to reserve their rights in certain circumstances, but an opt-out is only meaningful if it is technically usable, legally clear and respected throughout a complicated supply chain. A buried website instruction or a machine-readable marker offers limited protection when material has already been copied into datasets, redistributed through repositories, or used by multiple model developers.
Remuneration: permission without payment is not a fair market
Even where licensing is possible, creators need meaningful remuneration. AI firms can derive substantial commercial value from systems trained on vast quantities of books, journalism, lyrics, images, performances and audio recordings. If creators receive nothing, the result is a transfer of value from cultural workers to AI infrastructure owners.
Fair payment does not necessarily mean every author must negotiate a separate deal with every company. That approach would be unrealistic for both creators and users. Collective management, standard licences and representative agreements may offer more practical routes, provided they are transparent and distribute income fairly.
This question is acute for performers. A voice actor may be hired for one recording session, only to discover later that a synthetic voice can reproduce elements of their performance indefinitely. Contract terms must clearly distinguish a limited recording use from permission to create, train or deploy a digital replica. A one-off fee should not silently become an unlimited buyout of a person’s voice, style or professional identity.
Transparency is the enabling condition for authorisation and remuneration. Creators cannot license, reserve rights or pursue payment if they cannot identify the works used, the datasets involved, the model provider, and the commercial uses that follow.
The EU AI Act includes transparency obligations for providers of general-purpose AI models, including requirements related to copyright compliance and publicly available summaries of training content. The crucial policy battle is now about implementation and enforcement. A vague description such as “internet data” does not allow an author, publisher, collective management organisation or performer to assess use. Useful transparency needs to be sufficiently specific, auditable and accessible to rights holders.
Why CULT Committee scrutiny matters
The CULT Committee does not operate in isolation: AI governance also touches the internal market, legal affairs, employment, competition and data policy. But culture cannot be treated as a side issue in a technical regulation file.
Europe’s cultural sector is unusually diverse. It includes many small publishers, independent labels, literary translators, public-service media organisations, theatre companies and freelance performers working across languages with relatively small markets. A model in which English-language, high-volume online material is scraped and monetised at scale can further disadvantage minority-language and local European creation.
That has implications beyond individual income. If professional authors and performers cannot sustain careers, fewer works will be commissioned, translated, recorded and distributed. The long-term cost is not just lost royalties; it is less cultural plurality and greater dependence on content shaped by the priorities of dominant platforms.
The joint letter therefore frames copyright and performers’ rights as conditions for cultural independence. Europe cannot credibly support linguistic diversity, independent publishing and local production while allowing the underlying creative catalogue to be appropriated without reliable consent or compensation.
What creators and cultural businesses can do now
Review contracts before accepting AI clauses
Authors, translators, performers and freelancers should examine contracts for broad language covering “new technologies,” “all media now known or later developed,” training data, synthetic outputs or digital replicas. Such wording can grant far more than the immediate project requires.
Ask for specific answers: Is AI training permitted? Which model or provider is involved? Is the permission exclusive? How long does it last? Can it be withdrawn? Is there separate payment? Can the work or voice be used to generate new material? Where possible, seek legal advice through a union, professional association or rights organisation before signing.
Maintain dated copies of manuscripts, recordings, contracts, metadata and registration records. For publishers and producers, clean rights documentation is essential. It helps establish who can license a work and supports future claims if unauthorised use is suspected.
Creators should also ensure that websites, portfolios and distribution platforms carry accurate copyright and contact information. This is not a complete defence against scraping, but it strengthens the evidence trail and makes legitimate licensing easier.
Support collective solutions, not only individual opt-outs
Individual creators should not be expected to police global AI markets alone. Joining a writers’ association, performers’ union or collective management body can increase bargaining power and improve access to contract guidance.
For European cultural businesses, the constructive alternative is not to reject AI outright. It is to build licensed datasets, negotiate representative agreements, document provenance and pay creators when their work generates value. Responsible AI procurement should include a basic question: can the supplier explain where its training material came from and how rights were cleared?
The policy test: enforceable rights rather than voluntary promises
The ART principle is ultimately a test of whether AI policy produces enforceable rights. A voluntary pledge to “respect creators” is insufficient if there is no usable licensing route, no dataset disclosure, no audit mechanism and no remedy when rights are ignored.
European lawmakers should focus on implementation details: robust and standardised training-data documentation; effective ways to express and honour rights reservations; collective licensing mechanisms that work across borders; protections against abusive contract terms; and meaningful sanctions for non-compliance. They should also ensure that transparency rules do not become so general that they conceal rather than reveal the origin of training material.
The message from the 14 federations is timely because the decisive choices are no longer abstract. AI systems are already changing commissioning, search, translation, narration, advertising and content production. Preserving authorisation, remuneration and transparency is not an attempt to freeze technology. It is the minimum framework needed to ensure that innovation does not depend on uncompensated cultural labour.
FAQ
What is the ART principle in relation to AI?
ART means Authorisation, Remuneration and Transparency. It calls for creators’ permission before protected works are used, fair payment where AI businesses derive value from them, and clear information about training data and model use.
Does the ART principle mean creators want to ban generative AI?
No. The principle supports a rights-respecting AI economy. It argues that AI tools should be built and deployed through lawful licensing, transparent sourcing and fair contractual terms rather than through opaque extraction of creative work.
The answer depends on the type of work, where it is published, applicable copyright rules and the technical means available to reserve rights. Because these systems remain complex and inconsistent, creators should seek advice from their professional organisation or a qualified copyright lawyer rather than assume an opt-out has been implemented effectively.
Performers may face the reuse of recordings to create synthetic voices, likenesses or new performances. Clear, separate consent and appropriate compensation are essential because permission for one recording session should not automatically authorise unlimited digital replication.
Fuente: European Writers Council — Mon, 02 Mar 2026 08:00:00 GMT
