Skip to main content
Directive on copyright and related rights in the Digital Single Market (DSM Directive) (Directive (EU) 2019/790)

Status: In force

  • Since 6 June 2019
  • To be transposed into national law until 7 June 2021


The DSM Directive aims to modernize copyright law to adjust existing rules to the needs of a digital society in the 21st century. Most notably, the DSM-Directive foresees a new liability regime for Online Content Sharing Service Providers (“OCSSP”), a new neighbouring right for press publishers, new options for collective licensing as well as a text and data mining exception which may become relevant in the AI context.


While the Directive applies to all copyright holders and copyright users, most rules target specific groups. The most relevant changes apply to online platforms (OCSSP and Information Society Services (“ISS”)), press publishers, researchers/cultural heritage institutions and collective management organisations (“CMO”). The text and data mining provisions will be relevant for AI developers and right holders alike.

Key elements

  • OCSSP perform an own act of communication to the public regarding content uploaded by their users. They must make best efforts to obtain authorisation for such use and to ensure a takedown/staydown of non-authorized content once sufficient information is provided by rightsholders;
  • Press publications are protected by a new neighbouring right. The neighbouring right applies to uses by ISS, but excludes private and non-commercial uses by individual users, hyperlinking, the mere facts reported as well as the uses of individual words or very short extracts of publications;
  • Member states may introduce rules for “extended collective licensing” which allow CMOs – under narrow conditions – to issue licenses for the works of “outsiders” (i.e. rightsholders without a rights management agreement with the CMO);
  • New copyright exceptions for text and data mining – though relatively narrow in scope – may become relevant regarding the collection and use of training data in the AI context (especially for foundation models used for genAI applications).


  • Negotiation of adequate license agreements regarding use of works on OCSSP platforms as well as the balancing of rights (takedown/staydown vs. user rights) is and will remain difficult.
  • The scope of the press publisher rights will remain a source of legal uncertainty and heated discussions between press publishers and ISS for years to come.
  • The possibilities (and boundaries) to introduce extended collective licensing regimes to facilitate the rights clearance (“in bulk”) are only beginning to be explored.
  • When the DSM-Directive was drafted, the new, rapid developments in the AI context were not entirely foreseeable. It remains to be seen whether the new exceptions for text and data mining can provide answers for some questions, nonetheless.
  • Overlap with DSA.