Remuneration of Cultural Content Used by AI – Report by Two CES Researchers for the CSPLA.

The CSPLA Publishes the Mission Report on the Remuneration of Cultural Content Used by Artificial Intelligence Systems

The mission led by Professors Alexandra Bensamoun and Joëlle Farchy has released its report on the remuneration of cultural content used by AI systems. The analysis is presented in two parts: one legal, the other economic.

Alexandra Bensamoun, Professor of Law at the University of Paris-Saclay and former member of the Artificial Intelligence Commission, and Joëlle Farchy, Economist and Professor at Paris I Panthéon–Sorbonne, have published their report on AI and the remuneration of cultural content. The mission was carried out with the support of two co-rapporteurs: Julie Groffe-Charrier, Associate Professor (HDR) in Law, and Bastien Blain, Junior Professor at Paris I Panthéon-Sorbonne.

Context

Machine learning technologies, which lie at the heart of artificial intelligence (AI) systems, “train” on vast amounts of cultural data.

With regard to data protected by literary and artistic property rights, French rights holders, faced with the growing use of their works and performances, are increasingly choosing to exercise their right to object (“opt out”). While this constitutes a significant step forward in protecting copyright in a rapidly evolving context, opt-out remains insufficient. If applied on a large scale, it could undermine both the reliability of AI outputs and the visibility of French content.

The principle of source transparency will enable rights holders to:

verify that their opposition to the use of their works and performances has been respected;

draw the corresponding economic consequences in terms of remuneration.

From an economic perspective, access to cultural data is a key issue to ensure that business models and value chains are respected. Since large-scale data scraping is likely to disrupt the balance of the sectors concerned, their monetization must be ensured with a view to the sustainability of all stakeholders’ business models, while also respecting public policy objectives such as cultural diversity.

From a legal perspective, the conditions now appear to be in place for the emergence of a balanced data market—one that ensures both fair remuneration for rights holders and legal certainty for AI model providers. Negotiating a license or paying remuneration to rights holders must therefore be aligned both with the European legal framework and with the economic feasibility of the solutions proposed.