Executive Summary – Global Digital Encounter 39: Generative AI Models at the Gate – Licensing Frameworks for the Effective and Efficient Protection of Copyrighted Content in an AI World

September 10, 2025 | 4:00–5:00 PM CEST

The 39th Global Digital Encounter (GDE) brought together distinguished experts to tackle one of the most pressing challenges at the crossroads of technology, law, and economics: how to design licensing frameworks that protect copyright while enabling the training of generative AI models. With a focus on the music industry, but with broader implications for all copyright-based sectors, the session examined whether voluntary bilateral licensing or collective licensing systems provide the fairest and most efficient balance between creators’ rights and developers’ needs.

This discussion was moderated by Prof. Laurent Manderieux (Bocconi University, Director of TIPSA, Co-founder of the GDE), who emphasized from the outset that the rise of Generative AI has created a “cutting-edge issue” for the IP community—one where exclusive rights, licensing models, and innovation incentives collide.

Objectives

• Assess whether current copyright licensing models are adequate for AI training.
• Explore voluntary vs. mandatory licensing, and bilateral vs. collective frameworks.
• Address transparency challenges in AI training datasets and outputs.
• Examine the role of remuneration, transaction costs, and moral rights in future frameworks.
• Consider the geopolitical context shaping transatlantic and global regulatory choices.

Panel

• Jorge Padilla (Spain) – Board Member of Fundación Fide; Chair of its International Academic Council; Senior Managing Director at Compass Lexecon.
• Prof. Javier Fernández-Lasquetty (Spain) – Professor of Intellectual Property Law at IE University; IP/IT Of Counsel at Ramón & Cajal.

Moderator: Prof. Laurent Manderieux (Italy) – Professor of IP Law, Bocconi University; Director of TIPSA; Co-Founder of the GDE.

Key Discussions

1. The Disruption of Traditional Licensing Frameworks

• Javier Fernández-Lasquetty highlighted the radical difference between traditional copyright licensing (covering uses like reading a book or streaming music) and AI training, which involves mass ingestion of countless works through opaque “black-box” processes.
• He noted unresolved questions of authorship, liability, and copyright infringement in AI-generated outputs, warning that mass litigation in areas such as music plagiarism could follow.

2. Economic and Industry Perspectives

• Jorge Padilla, drawing on his recent work for the music sector, explained how AI threatens to cannibalize revenues of creators whose works are used for training but not compensated.
• He stressed two critical challenges:
  1. Transparency – it is nearly impossible to detect what copyrighted works were used in AI training.
  2. Remuneration – copyright holders face undercompensation while AI developers profit from outputs trained on their content.
• Padilla underscored the parallel with standard-essential patents (SEPs), where mandatory licensing is justified only when content is truly indispensable. For most copyrighted content, however, voluntary licensing is preferable.

3. Voluntary vs. Mandatory, Bilateral vs. Collective Licensing

• Padilla’s position: voluntary licensing should prevail, with freedom for rightsholders to choose bilateral agreements (as majors already do with streaming platforms) or collective licensing (supporting smaller creators).
• He cautioned that mandatory collective licensing could create monopolistic structures, eliminate competition, and lead to regulatory disputes over pricing.
• Fernández-Lasquetty broadly agreed but stressed the need for collective solutions for small creators who lack bargaining power. He also insisted on preserving moral rights—the right of authors to refuse uses they oppose, even if economically lucrative.

4. Output Liability and Human Creativity

• Fernández-Lasquetty emphasized that licensing contracts must address liability for infringing outputs, a dimension absent from traditional copyright contracts.
• He also pointed out that many authors might oppose the use of their works for AI training on ethical or artistic grounds, not just economic ones, raising the importance of protecting human choice and creative integrity.

5. Regulatory and Geopolitical Dimensions

• Padilla warned that transatlantic dynamics could reshape the debate:
  • In the US, a split exists between strong pro-IP companies and Big Tech, which favors weaker copyright rules to accelerate AI development.
  • Europe’s stricter copyright stance risks being seen by the US as hindering its ability to compete with China in the AI race.
• Fernández-Lasquetty added that the US government and industry see AI as another “Sputnik moment”, where being first is imperative, even if it means challenging traditional copyright norms.

Conclusion

The session concluded that licensing frameworks must evolve to strike a careful balance:

• Voluntary bilateral licensing remains the most efficient system for large players.
• Collective licensing pools are vital to empower smaller creators.
• Transparency requirements and clear liability rules must underpin any system.
• Moral rights and human creativity should not be sidelined in the rush to regulate.

Prof. Laurent Manderieux closed by stressing that this debate is not only about music or copyright but about reshaping the elasticity of the entire IP system in response to technological disruption. The GDE will continue to serve as a platform to track and shape this transformation.