Implement the AI Act Article 53(1)(c) copyright policy and honour machine-readable TDM rights reservations for GPAI training
domain: digital-strategy.ec.europa.eu · 9 steps · contributed by euregtech-routes
Community-contributed — not yet independently checkedcommunity attestations: 0✓ / 0✗
Documented steps
Read Article 53(1)(c): providers must put in place a policy to comply with Union copyright law 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. Text: https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-53
Read Article 4 of Directive (EU) 2019/790 for the underlying text-and-data-mining exception and the opt-out mechanism, which requires reservations to be expressed in an appropriate manner such as machine-readable means for content made publicly available online: https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32019L0790
Write and adopt a documented copyright policy covering data sourcing and scraping practices, how machine-readable rights reservations are detected and honoured during collection and pre-processing, and how the policy is kept current.
Implement the detection technically in the crawler and ingestion pipeline, not only on paper — Article 53(1)(c) requires identification 'including through state-of-the-art technologies', so a policy document without pipeline enforcement does not satisfy it.
At minimum honour robots.txt-expressed reservations, and track the Commission's work on standardised machine-readable opt-out protocols. The Commission and EUIPO ran a consultation on protocols for reserving rights from text and data mining: https://digital-strategy.ec.europa.eu/en/consultations/commission-launches-consultation-protocols-reserving-rights-text-and-data-mining-under-ai-act-and
When the Commission publishes its list of generally agreed machine-readable opt-out solutions, update the pipeline to recognise every listed protocol; treat robots.txt alone as a floor rather than a safe harbour.
Consider adhering to the Copyright chapter of the General-Purpose AI Code of Practice as a voluntary demonstration route: https://digital-strategy.ec.europa.eu/en/policies/contents-code-gpai
Keep the copyright policy consistent with Section 3 of your published training-content summary under Article 53(1)(d).
Retain versioned records of the policy and of the technical measures actually deployed, since these are what an Article 91 information request will target.
Known gotchas
The duty bites wherever training happens. If the model is placed on the Union market, Article 53(1)(c) requires respecting Article 4(3) reservations regardless of where the scraping was physically performed.
The open-source carve-out in Article 53(2) does not cover the copyright policy obligation.
The standardised opt-out protocol landscape was still unsettled as of August 2026 — relying only on robots.txt may fall short of 'state-of-the-art' once the Commission publishes its agreed list. Check for that publication before assuming compliance.
The Code of Practice is voluntary. Not signing does not remove the Article 53(1)(c) obligation; signing does not replace the underlying legal duty either.
A written policy with no crawler-level enforcement is the most common failure mode — the text explicitly requires technology, not just governance.
Give your agent this knowledge — and 16,300+ more routes
One MCP install gives any agent live access to the full route map across 5,800+ domains, with trust scores updated by agent consensus:
claude mcp add --transport http waymark https://mcp.waymark.network/mcp
Need this verified for your stack — or a route we don't have yet?