Some fifteen European organisations representing journalists, literary translators, scriptwriters and songwriters urged the co-legislators, on Thursday 23 November, to “put transparency back at the heart of the EU Artificial Intelligence Act” (AI)’s interinstitutional negotiations (‘trilogues’) (see EUROPE 13297/24).
In particular, the signatories want the text to oblige entities deploying AI systems to prove that the training of their AI has been carried out in compliance with applicable European and national law, in terms of “intellectual property, the protection of personal data or other relevant provisions”.
“None of the protections provided by the GDPR (General Data Protection Regulation, editor’s note) and the CDSM (Copyright in the Digital Single Market Directive) have the slightest chance to work if appropriate transparency requirements, including strong record-keeping and transparency obligations regarding the use of copyright-protected content by generative AI models, are not included in the EU AI Act”, they insist.
In addition, the organisations want strict “visible and/or audible” content labelling obligations to be imposed on all deployers of generative-AI powered technologies.
“While these obligations may be adapted to the nature of the content in order not to hinder its exploitation, we firmly reject broad exceptions that would render labelling obligations meaningless in practice”, the signatories state.
See the document: https://aeur.eu/f/9po (Original version in French by Thomas Mangin)