AI Act – GESAC welcomes today’s vote at the European Parliament in favour of including Generative AI systems and requiring them to respect copyright

Today the ITRE and LIBE, the leading Committees on AI Act adopted the European Parliament’s position for the trialogue negotiations with the Council and the European Commission. Considering the latest developments in the “Generative AI” technologies and foundation models that deeply impact the society and raise serious threats on fundamental rights and European values, MEPs rightly decided in the adopted report to include certain obligations on the development and operation of such technologies in the scope of this much needed European AI regulation. In particular, the adopted report is requesting generative AI technologies and foundation models to comply with the Union law and to provide transparency on their training data, which might include copyright protected works, when they generate content such as complex text, images, audio or video. The report is still to be approved in plenary.

Véronique Desbrosses, General Manager of GESAC said “Artificial intelligence brings about many new opportunities for our lives, including in the field of culture and creativity, but its applications also raise a number of ethical and legal questions due to their wide-spread uncontrolled and unchecked usage. Transparency and accountability standards are necessary as regards data sets used to train AI tools that generate new content based on preexisting works of authors. We welcome the positive direction taken in the European Parliament and trust that this will be translated into a clear and effective obligation in the final version of the AI Act.”

She added “Transparency obligation on foundation models and generative AI systems is key both for the public to understand how a content is generated by AI and for creators to see if their protected works are used without authorisation. AI tools can significantly help authors to create new artistic works with more opportunities and efficiency, but scraping of their works without consent and remuneration to create a new and parallel market of humanless content would be devastating.”