ľ�ϸ���Ӱ��

Article 3 of the EU Directive on Copyright in the Digital Single Market (DSM), and the national implementations thereof, research organisations such as the UU, are allowed to carry out text and data mining activities (TDM) for the purposes of scientific research, as long as they have lawful access to the (copyright or databases right protected) works used as input.

TDM is defined as “any automated analytical technique aimed at analysing text and data in digital form in order to generate information which includes but is not limited to patterns, trends and correlations” When these conditions are met, AI tools might be used. Thus, it is likely that subscription databases (to which the UU has lawful access) may be used with AI for research purposes. Besides Article 3, there is a broader exception under Article 4 DSM that allows TDM for any purpose, including commercial uses - but this applies only if the content you're mining hasn't been explicitly reserved by the copyright holder (an "opt-out"). Also, both exceptions require that you have lawful access to the material (e.g., via a library subscription or legitimate purchase), meaning you can’t just use freely available or pirated content.

In short, if you're using copyrighted content with an AI tool just to help you work or the study of your students, the TDM exceptions may not always allow you to feed copyright protected materials in an AI-tool. In those cases: always check the terms of use for the content and consider whether the rights holder allows such use.

The following rules apply to the use of copyrighted material by teachers and students within AI platforms. This ensures compliance with Dutch copyright law and any existing license agreements with publishers.

General principles

• Works from publishers such as Pearson etc. may be uploaded to the AI platform insofar as this is permitted under the terms of use and license conditions and any additional agreement concluded with UU in this regard, or for which permission has been obtained.
• No complete copyright-protected works may be uploaded or shared on the AI platform without the permission of the author(s) of the work in question.
• Do not upload complete works (e.g. entire books or articles behind paywalls) that have been obtained illegally.
• Preferably, do not upload complete books, articles or databases, but refer to them via links (to DOI, library access, publisher portal, etc.).
• Check whether material is available under open access or a Creative Commons license.
• The use of open access and open educational resources is preferable (where possible).
• Works produced by UU itself (such as readers) must have names of students, lecturers or other personal data removed, prior to uploading (or pasting/copying into the platform), unless the authors have no objection to this and have communicated this personally in one-to-one contact.

Educational exceptions

• Educational exceptions (Section 16 of the Copyright Act) only permit the use of limited excerpts for educational purposes and with source attribution.
• Use excerpts from texts, images or tables comprising a maximum of 20% of the work or 40 A4 pages.
• Licenses and collective agreements (e.g. Stichting PRO, UvO, library licences) take precedence over ad hoc use.
• There is no maximum for images in publications or presentations (PowerPoint, etc.). However, there are two conditions: a maximum of 10 works by the same creator and no more than 25 images from the same work.
• More information can be found at:

Remarks for teachers

Inform students explicitly about what may and may not be uploaded or shared onto the AI platform.

For example, upload a chapter or paragraph instead of an entire book from a publisher that falls under the educational exception.

In exceptional cases

Notice-and-takedown procedure: upon notification of infringement, the material can be removed from the AI platform immediately.

Copyright originates (throughout the European Union) only when a work is the result of the author’s own intellectual creation. The so-called anthropocentric requirement demands that it is a human who makes the creative choices necessary for copyright to come into existence. Thus, based on current EU copyright law, as GenAI-output lacks “human creativity”, there will be no copyright protection for results such as AI-generated lecture-slides, assessment questions or visuals. That might change when a human (you, for instance) makes changes to such output, as long as these changes in turn can be qualified as the author’s own intellectual creation. In such cases, it is either you, or (likely) the university, as your employer, who will hold the copyrights thereto.

In principle, all three can be considered to hold a degree of liability but what is important for you is that you have a duty of care to ensure that the output does not violate copyright of others.