General-Purpose AI: Copyright Policy and Training-Data Summary (EU AI Act Article 53)

Article 53 obligations apply to any GPAI model provider placing a model on the EU market. The training-data summary is the first EU-wide transparency requirement covering pre-training corpora. Non-compliance can be enforced by the AI Office and national authorities.

Vendors in this space provide training-corpus provenance tracking, rights-reservation detection (including honoring TDM opt-outs expressed under Directive 2019/790 Article 4(3)), training-data disclosure templates aligned to the AI Office template, and technical documentation aligned to Annex XI.

• Maintain up-to-date technical documentation of the model per Annex XI (training/testing process, evaluation results)
• Maintain documentation for downstream AI system providers integrating the model (Annex XII)
• Implement a policy to comply with Union copyright and related-rights law
• Identify and honor rights reservations expressed under Directive (EU) 2019/790 Article 4(3), including through state-of-the-art technologies
• Publish a sufficiently detailed summary of training content using the AI Office template
• Cooperate with the European Commission and national competent authorities on request
• Where applicable, follow codes of practice under Article 56 or European harmonised standards, or demonstrate equivalent means of compliance

Article 53(1)(a) and (b) do not apply to models released under qualifying free and open-source licenses, EXCEPT for GPAI models with systemic risk (Article 51). The copyright policy and training-data summary obligations in Article 53(1)(c) and (d) apply regardless of open-source status.

Source: Article 53 — Obligations for Providers of General-Purpose AI Models, EU Artificial Intelligence Act (artificialintelligenceact.eu). Applies from 2 August 2025 per EU AI Act Article 113.