EU proposals to ensure transparency and fair remuneration for use of copyrighted work by GenAI

The question

What is the EU doing to protect the rights of copyright owners against GenAI?

The key takeaway

A draft report by Members of the European Parliament's Legal Affairs Committee (LAC) has identified significant legal uncertainty in EU copyright rules, including the Digital Single Market Directive, which were not designed to regulate AI training. It highlights unresolved questions over compensation for copyright holders whose works are used to train GenAI, calls for AI providers to disclose the copyrighted content they use for training, and emphasises the need for clear rules on what constitutes lawful use, opt-out mechanisms, and enforcement of rights. Ultimately, the European Parliament aims to encourage AI innovation and development while safeguarding Europe's creative and cultural sectors through IP protection.

The background

On 28 January, the LAC approved new proposals aimed at clarifying how EU copyright law should apply to GenAI operating in the EU market. The proposals, which were ratified by a majority, respond to growing concerns that GenAI models rely heavily on copyrighted works for training often without transparency, consent, and duly compensating creators. The draft LAC report is being finalised and put to a Parliamentary vote in plenary in March 2026. The report seeks to ensure that access to high quality data for AI development works concurrently with fair remuneration and legal certainty for the creative sector.

The development

The report contains proposals for the European Commission and other EU institutions, and it calls on the Commission to immediately evaluate whether existing copyright laws are fit for the legal challenges posed by GenAI.

• Core principle: at the core of the proposals is the principle that EU copyright law must apply to all GenAI systems made available in the EU, regardless of where training occurs.
• Transparency obligations: AI providers must:
  • produce detailed records of web crawling activities
  • disclose which copyrighted works are used in training datasets. Non-compliance could amount to a copyright infringement and where providers fail to properly disclose, they would be obliged to prove that no copyrighted material was used.
• Compensation: the LAC asks the Commission to consider whether compensation mechanisms for rights holders should apply retroactively.
• Licensing: to support lawful access to content, LAC proposes:
  • onew licensing rules
  • ovoluntary collective licensing schemes accessible to creators of all sizes, ensuring work is protected across the creative spectrum.
• Opt-out mechanisms: the LAC calls on the Commission to develop tools that enable rightsholders to protect their work from general-purpose AI systems, including standardized machine-readable opt-out mechanisms, potentially managed by the European Union Intellectual Property Office (EUIPO).

The LAC does not support a global licence scheme which would allow AI providers to train models in exchange for a flat-rate payment, and also opposes granting copyright protection to content generated entirely by AI, calling instead for safeguards against manipulated, misleading or illegal AI-generated content.

Finally, the LAC stresses concerns with the news media sector, warning that selective aggregation of news content by AI systems risks undermining media pluralism which can reduce traffic, revenues, and trust. It therefore supports media autonomy when deciding whether they consent to their content being used for training purposes.

Why is this important?

GenAI is rapidly developing and as important as it is for the EU to remain globally competitive in AI technology, the fundamental rights of the creative industry should not be neglected. Compensating rights holders when their copyrighted work is used by GenAI is important to ensure the economic basis for creative industries is not undermined. It is clear that a fully developed and enforceable legal framework tackling AI could be the key to ensuring the EU becomes a global AI pioneer in a safe, predictable environment thus strengthening its market position.

Any practical tips?

Rights holders and creators should be aware of any opt-out mechanisms and licensing arrangements that will ensure the protection of their work. They should also document any current use of their work by AI datasets in order to enforce their rights immediately when legal frameworks are solidified.

On the other hand, AI developers and tech companies should monitor developments in EU law on AI training and it is likely to be helpful to maintain a clear and transparent record of what copyrighted material they use for training, so they are not caught off guard when the new measures are enforced. If the report is approved in the Parliamentary vote, they should also seek advice from legal experts on how to prepare thoroughly for their forthcoming obligations and should ensure they have enough cashflow for potentially remunerating rights holders when the time comes.

Spring 2026