TLDR: A recent study commissioned by the European Parliament’s Committee on Legal Affairs (JURI) has concluded that current EU copyright laws, particularly text-and-data-mining (TDM) exceptions, are inadequate for the extensive training requirements of generative AI. Published on July 9, 2025, the study advocates for a significant overhaul of the existing ‘opt-out’ regime, proposing a shift towards frameworks that prioritize explicit consent, fair remuneration for creators, and increased transparency in AI training data usage. This signals a growing momentum within the European Parliament to modernize copyright rules to address the challenges posed by advanced AI technologies.
On July 9, 2025, the European Parliament’s Committee on Legal Affairs (JURI) released a pivotal study examining the intricate relationship between generative artificial intelligence (AI) and European Union copyright law. The study, titled ‘Generative AI and Copyright: Training, Creation, Regulation’ and authored by Nicola Lucchi, PhD, asserts that the current legal framework, specifically the text-and-data-mining (TDM) exceptions outlined in Directive (EU) 2019/790 on Copyright in the Digital Single Market (CDSM Directive), is fundamentally ill-equipped to handle the vast scale of data required for training modern generative AI systems.
The core issue identified by the study is the existing ‘opt-out’ mechanism under Article 4 of the CDSM Directive. This system, which presumes the lawfulness of copying unless rightsholders proactively embed machine-readable reservations (such as robots.txt), ‘effectively treats silence as consent,’ according to the report. This approach is deemed fragmented, technically fragile, and largely unenforceable, raising serious doubts about its effectiveness as a reliable safeguard for rights holders. The study argues that these TDM exceptions were not designed for the nature and scale of generative AI training, making their current application highly contested.
In response to these challenges, the study strongly urges lawmakers to revise the framework to emphasize consent, remuneration, and transparency. Key recommendations include a potential shift from an ‘opt-out’ to an ‘opt-in’ model, which would require explicit permission from copyright holders for their works to be used in AI training. Furthermore, it proposes the establishment of collective licensing models and explores the possibility of extending copyright protection to certain AI-assisted creations. The report also calls for clear rules on input/output distinctions and harmonized opt-out mechanisms.
Also Read:
- Atlassian Co-founder Scott Farquhar Advocates for AI Copyright Law Reform in Australia Amidst Heated Debate
- Survey Reveals Over Half of Journalists Now Employ Generative AI
This comprehensive analysis underscores the Parliament’s readiness to adapt copyright rules to the evolving landscape of generative AI. It highlights the legal mismatch between AI training practices and current exceptions, as well as the uncertain legal status of AI-generated content. These developments pose structural risks for the future of creativity in Europe, where a rich and diverse cultural heritage relies on the continued protection and fair remuneration of authors. The study advocates for a ‘balanced regulatory model’ that does not treat innovation and authorship as opposing values, aiming to promote technological advancement while safeguarding creators’ rights. Developers, publishers, and other rightsholders are advised to closely monitor upcoming debates and prepare for potential shifts in licensing, data governance, and compliance practices.
