Yesterday, we launched our Guidelines for the Implementation of the DSM Directive. This is the first of a series of blogposts dedicated to the various provisions analysed in our guidelines. Today we give a quick explanation of the two mandatory exceptions for text and data mining contained in the new Copyright Directive.
For a detailed analysis, please read LIBER and Communia’s guide on Articles 3 and 4, authored by Benjamin White and Maja Bogataj Jančič.
What is at issue in Articles 3 and 4?
Although Big Data, AI and data mining is ubiquitous in the modern world, only a few European countries have limitations and exceptions which allow these activities where clearing rights is required but not appropriate or at all proportionate. These exceptions have been introduced to support innovation in modern data markets within the European Union.
Breaking down Articles 3 and 4
What can be done? Reproduction of copyright works and other subject matter and extraction and/or re-utilization of the whole or of a substantial part of the contents of a work subject to the sui generis database right.
For which purposes? Article 3: Text and data mining for scientific purposes, including most commercial ones // Article 4: Text and data mining for any purpose, including commercial ones
What content can be used? Article 3: Works and other subject matter, including databases but prima facie excluding software // Article 4: Works and other subject matter, including databases and software
Who can make those uses? Article 3: Anyone who has lawful access to content via a cultural heritage institution, educational establishment or a research organisation // Article 4: Anyone who has lawful access to content and who is not prevented from data mining by contract, or if on the open web, by machine readable expressions of terms of use.e.g. robots.txt protocol.
For free or paid? Article 3: For free // Article 4: Unclear
How long can the derived data be stored? Article 3: No limits // Article 4: As long as necessary for the purposes of text and data mining
Are the exceptions protected against contract and technological override? Yes
How to deal with Articles 3 and 4?
Implement Articles 3 and 4. In countries where there is no political willingness to take full advantage of the policy space that is available under the existing EU laws, one should advocate for the best possible implementation of Articles 3 and 4.
That means for Article 3:
- advocating for a 72 hour response time in law where access is being blocked;
- rejecting specific requirements around secure storage and third parties holding data.
And that means for Article 4:
- advocating for a 72 hour response time in law where access is being blocked;
- rejecting any specific limitations on being able to retain data;
- rejecting any calls for the exception to be subject to remuneration;
- requiring that the Robot Exclusion Standard is used to disallow data mining on websites accessible on the open web.
Move beyond Articles 3 and 4. Member States can adopt text and data mining exceptions that are broader in scope than Articles 3 and 4 and yet are compatible with EU law. The ideal scenario would be to improve the legal framework in each country.
That means for Article 3 (in addition to the asks above):
- ensuring that the mining of software is expressly permitted in national law;
- requiring remote access in national law when data mining relates to the digitisation of analogue items
- advocate for the introduction of a clear exception for sharing the results of data mining.
And that means for Article 4 (in addition to the asks above):
- requiring remote access in national law when data mining relates to the digitisation of analogue items;
- advocating for the introduction of a clear exception for sharing the results of data mining;
- calling for clear language in contracts when rightsholders wish to reserve rights in materials not available on the open web.
