This Wednesday, April 23, on the occasion of World Book and Copyright Day, Rachida Dati, Minister of Culture, and Clara Chappaz, Minister Delegate for AI and Digital Affairs, announced the launch of a consultation cycle between representatives of generative AI model developers and those of cultural and media rights holders. This initiative aims to establish a constructive dialogue reconciling technological innovation and copyright protection.
Since the emergence of GenAI, protests by authors, artists, and publishers have multiplied against the unauthorized use of their works for AI model training. Last month, a hundred authors demonstrated in front of Meta's London headquarters to denounce the exploitation of their works in violation of their copyright. According to the government, this consultation cycle "should facilitate the understanding of mutual issues between AI developers and rights holders, highlight common interests, and identify best practices".

A Framework for Ethical and Respectful AI Creation

Building on the dialogue dynamics between these two categories of stakeholders initiated during the AI Action Summit last February, the consultation cycle will rely on the work of the Higher Council of Literary and Artistic Property (CSPLA). According to the ministers, it should "promote better valorization of cultural data," through the negotiation of explicit licensing agreements, which are still rare in the sector. The improvement of opt-out mechanisms, often deemed inadequate by rights holders, will also be addressed to give creators better control over the use of their works.
By promoting access to French and European cultural corpora under clear legal conditions, this initiative aims at developing competitive and sovereign generative AI models.
To lead this consultation cycle, the ministries have appointed two experts: Marc Bourreau, Doctor of Economics, Professor of Economics at Telecom-Paris (Institut Polytechnique de Paris), and Maxime Boutron, Master of Requests at the Council of State. The stakeholders, identified by the Ministry of Economy, Finance, and Industrial and Digital Sovereignty, and the Ministry of Culture, will be informed of the modalities and date of the first consultation meeting scheduled for spring 2025.

To better understand

What are the specific challenges associated with using copyrighted works to train AI models?

Using copyrighted works for AI training raises issues of reproduction and distribution rights, as well as fair compensation for creators. Rightsholders seek control and remuneration mechanisms like explicit licensing agreements.

How can explicit licensing agreements enhance the governance of cultural data?

Explicit licensing agreements set clear and legally binding conditions for the use of protected works, promoting fair compensation for creators and responsible use of cultural data. They also enhance transparency and accountability for AI developers.