It's a similar feeling to looking at a blurry photo: in the last three years, generative artificial intelligence (GAI) has flooded almost every area of daily life and does not seem to be receding in the immediate future. We don't quite know why, but it's there and it's changing things at a dizzying pace. Between anecdote and more rigorous application, there are numerous uses being given to it: from improving citizen attention by public administration to optimizing management and diagnostic processes in the health sector, passing, of course, through simple experimental and domestic play. It is evident, therefore, that artificial intelligence can be a useful tool to improve people's lives, but like any tool that is born and implemented in an ultra-fast and massive way, it is necessary that it be regulated and critically analyzed. Because, although we know or intuit its potential, what should be its limits? Where should artificial intelligence end and leave space for human sensitivity and reasoning? What implications does GAI have in the field of copyright and cultural rights in general?
Given this sudden irruption, numerous organizations and platforms have set to work to protect creators' rights against the misuse of GAI. The Associació d'Escriptors en Llengua Catalana (AELC), representing nearly 2,000 literary authors from all over the Catalan Countries, is one of them, and has been able to participate in part of the negotiation of the European Regulation on Artificial Intelligence through different international federations to which it belongs, such as the European Writers' Council (EWC), a platform that brings together 53 professional associations with nearly 250,000 professional authors, or the European Council of Literary Translators' Associations (CEATL), with 38 associations and 12,000 member translators.
In this sense, and beyond the negotiation and copyright defense work carried out by both federations, a very valuable contribution from the EWC should be highlighted, which has surely marked and will continue to mark the conversation on this topic: the ART Precept. ART is the acronym for the three pillars that should structure any regulation, use, or reflection made in the field of AIG if authors' rights are to be taken into account, and it means nothing other than "Authorization," "Remuneration," and "Transparency." The three concepts are basic notions found in international intellectual property legislation, which should prevail over any technology or reality that has emerged subsequently.
They imply such essential issues as that any use of a work, whether for training or for feeding AIG systems, must be clearly authorized and in writing by its owner; that all use of that work must be adequately and proportionally remunerated, following the principles of EU Directive 2019/790 on Copyright; and that, finally, clear information must be provided on the works used to feed and train AIG systems (input), as well as on the texts (output) that these generate. Put another way: if a work is used, it must always be with permission, it must be paid for, and, finally, it must be made known.
As we have said, however, the image is in motion. And when progress is made in one aspect, new ones arise that require not quick, but urgent solutions. AELC itself published, in autumn 2023, one of the first guides on AIG for authors in Europe, and I myself contributed, with an international group of experts, to the elaboration of an EWC document of Recommendations on AI for the European book sector, published in 2024. This federation, and also CEATL, have contributed to the negotiation of the AI Regulation approved last year, a minimum text that protects some human rights of the population (for example, through the prohibition of the use of biometric controls and facial recognition or certain measures in data protection), but which, unfortunately, has been and continues to be insufficient in terms of copyright.
A sample of this is that the European Commission's AI Office has already launched not one, but two different (and voluntary) ethical codes to complement its application: a general one, presented last July, with a very softened proposal, decidedly watered down by large technology companies during the participatory drafting process, and another that will specifically address transparency, regulated by Article 50 of the Regulation, which we are currently debating.
And in the meantime? In the meantime, it will be necessary to monitor these codes and the application of the Regulation in each state. Also, the report prepared by MEP Axel Voss, a member of the European Parliament's Committee on Legal Affairs, who last September brought joy to creators by recognizing, in a text that may condition the debate in the coming years, the need to remunerate the uses of all copyrighted works, and at the same time that the text and data mining exception, included in the aforementioned EU Directive 2019/790, is intended for research contexts and not for commercial uses, as developers advocate.
Everything has gone very fast, yes. A lot of work has been done and there is still much to do. Informing ourselves and training ourselves is the first step to ensure that a tool like GAI, which promises a revolution comparable to that of the Internet in the nineties, brings more progress than harm to society as a whole. Otherwise, we would be facing a very clear and surely unforgivable case of not-so-generative human stupidity, which perhaps should also be regulated. In the meantime, and each within their scope, we must continue working. Contrary to what is often said, he who does not move does not appear in the photo, and we should all appear in it. Smiling, well-portrayed, proud of what we will have protected and, at the same time, advanced.