AI rights in Europe
https://www.europarl.europa.eu/RegData/etudes/STUD/2025/774095/IUST_STU(2025)774095_EN.pdf
Generative AI & Copyright: The EU Pushes for an Unwaivable Right to Equitable Remuneration
The European Parliament has just released a comprehensive 175-page study on the impact of generative AI on copyright. Among its most notable recommendations is the creation of an unwaivable right to equitable remuneration for authors and rightsholders whose works are used to train AI systems.
📚 A New EU-Level Exception for AI Training
To address the scale and complexity of training generative AI systems, the report proposes a new statutory exception at the EU level, allowing copyrighted works to be used specifically for AI training purposes.
However, this exception would not come without conditions. It would be coupled with a mandatory remuneration mechanism—one that is unwaivable, ensuring that creators receive fair compensation regardless of contractual terms or licensing models. This represents a shift towards protecting creators’ economic participation in value chains driven by data.
⚖️ Addressing a Systemic Market Failure
The proposal is grounded in a clear diagnosis: the current legal and licensing framework is structurally unfit to handle the industrial scale and speed of AI training. Individual licensing is simply not feasible, and as a result, human creators are being systematically excluded from the value derived from their works.
Even collective licensing or opt-in systems have proven insufficient to correct this imbalance. This is not merely a question of fairness—it is a recognition of a deeper market failure, where creators lack bargaining power in the face of large-scale data extraction.
🧩 A Tool to Rebalance Power in the AI Era
By introducing an automatic right to remuneration, the EU aims to restore a minimum level of economic agency to authors and rightsholders, whose works form the foundation of AI systems that can now generate substitutive content at scale and near-zero marginal cost.
Importantly, the report clarifies that this long-term model does not undermine the opt-in principle. Rather, it acknowledges that the existing framework lacks the structural capacity for meaningful compliance at the scale required by generative AI.