EU_AI_ACT_GPAI_COPYRIGHT_TRANSPARENCY_001

rules/regulations/eu/eu_ai_act_gpai_copyright_transparency.yaml

Citation

Regulation (EU) 2024/1689, Article 53(1)(c): put in place a policy to comply with Union law on copyright and related rights, and in particular to identify and comply with, including through state-of-the-art technologies, a reservation of rights expressed pursuant to Article 4(3) of Directive (EU) 2019/790;

Condition

Requires providers of general-purpose AI models to put in place a policy to comply with Union law on copyright and related rights, including identifying and complying with rights reservations expressed under Article 4(3) of Directive (EU) 2019/790. Article 53(1)(c) of the EU AI Act.

Layer 1 detects

one sequence: train, fine-tune (also "fine tune", "finetune") or retrain, in any form, then on, with or using, then copyrighted, proprietary or licensed, then data, content, material, works, text or corpus, then without, lacking or no, then disclosure, summary, transparency, documentation or permission ("train on copyrighted data without permission"). The words disclosure, summary and transparency point at the public training-content summary of Article 53(1)(d), a separate duty with its own rule, rather than at the copyright policy of Article 53(1)(c). It matches wording, not meaning: it does not read whether a copyright policy exists or whether rights reservations under Article 4(3) of Directive (EU) 2019/790 are honoured, so it misses the practice told in other words ("We scraped every book on the shadow library into the training set and ignored the robots.txt opt-outs."), and it fires on text that only names the practice, such as a refusal. Whether a text does the practice is judged in Layer 2.

Condition type

regex; semantic condition in the file, not run by the engine

Status in the file

review