On 10 March 2026, the European Parliament adopted an important resolution addressing the opportunities and challenges posed by generative artificial intelligence (GenAI) for copyright law. This initiative reflects the rapid development of AI systems capable of generating text, images, music, video and code on the basis of vast datasets used during the training phase.
While the European Union seeks to foster innovation and maintain technological competitiveness in the global race for artificial intelligence, the Parliament emphasises that the development of these technologies must remain fully consistent with the protection of intellectual property rights and the sustainability of the cultural and creative sectors.
The resolution first recalls that intellectual property constitutes a fundamental right within the European legal order, notably under Article 17 of the Charter of Fundamental Rights of the European Union. This protection must be guaranteed throughout the entire lifecycle of artificial intelligence systems, including their development, training and deployment.
Generative AI models are typically trained on large-scale datasets composed of pre-existing content, a significant portion of which may be protected by copyright or related rights. Through statistical learning, these systems detect patterns and structures in the training data and subsequently generate outputs that may mimic or reproduce elements of human creativity.
This technological architecture raises several complex legal questions. Among the most pressing are the legality of using copyrighted works in training datasets, the legal status of AI-generated outputs, and the mechanisms required to ensure transparency and fair remuneration for creators when their works are used in AI systems.
The Parliament further notes growing evidence of practices that may infringe copyright rules, including the large-scale scraping of protected works from the internet without authorisation, the use of pirated sources and the failure to comply with rights holders’ reservations against text and data mining. Such practices are seen as potentially undermining the economic foundations of the European creative industries.
Against this background, the Parliament considers that the current European copyright framework does not yet provide sufficient clarity for the training of generative AI models. It therefore calls on the European Commission to clarify the application of the Directive on Copyright in the Digital Single Market (CDSM Directive) to generative AI training practices and to assess whether additional regulatory measures may be necessary.
Several core principles are emphasised in the resolution.
First, transparency regarding training data is regarded as a key requirement. AI developers placing general-purpose models on the EU market should provide clear documentation concerning the copyrighted content used during training, thereby enabling rights holders to identify and enforce their rights where necessary.
Second, the Parliament stresses the importance of effective opt-out mechanisms allowing rights holders to exclude their works from AI training. Such mechanisms should rely on standardised machine-readable formats and may be coordinated by trusted intermediaries in order to ensure both legal certainty and practical usability.
Third, the resolution encourages the development of functioning licensing markets between AI developers and rights holders. Voluntary collective licensing schemes could play a significant role in facilitating lawful access to high-quality training data while ensuring that creators receive appropriate remuneration for the exploitation of their works.
In this context, the European Union Intellectual Property Office (EUIPO) is envisaged as a possible trusted intermediary, capable of facilitating transparency, managing opt-out registries and supporting the development of licensing solutions between the AI sector and creative industries.
Beyond purely legal questions, the resolution emphasises the broader economic and societal implications of generative AI. Cultural and creative industries represent a significant share of the European economy, accounting for approximately 4% of EU value added and employing around eight million people. The large-scale use of copyrighted works by AI systems therefore raises concerns regarding the sustainability of this ecosystem if such uses are not properly regulated and remunerated.
The Parliament also warns against the potential substitution of human creative output by mass-produced AI-generated content, which could gradually erode incentives for human creation and negatively affect the quality and diversity of cultural production.
To address these risks, the resolution proposes several key orientations. EU copyright law should apply to all AI systems made available on the EU market, regardless of where the training activities took place. Moreover, where AI developers fail to comply with transparency obligations, a rebuttable presumption could arise that protected works were used in the training process, thereby facilitating enforcement actions by rights holders.
Finally, the Parliament reiterates that, under current EU copyright principles and the case law of the Court of Justice of the European Union, copyright protection remains grounded in human authorship. Consequently, content generated entirely by AI systems without human creative contribution should not qualify for copyright protection and should in principle remain within the public domain.
Conclusion
Through this resolution, the European Parliament seeks to establish a balanced framework reconciling technological innovation with the protection of creative works. Rather than limiting the development of generative AI, the Parliament advocates for a regulatory environment that promotes transparency, supports the emergence of licensing markets and guarantees fair remuneration for creators.
The resolution thus represents a significant milestone in the ongoing evolution of European intellectual property law in the age of artificial intelligence and signals the likelihood of future legislative developments aimed at adapting copyright rules to the realities of generative AI technologies.
