Many critical assessments and amendment proposals have been presented concerning the draft EU Regulation on preventing the dissemination of terrorist content online. European Parliament is under considerable time pressure to adopt the regulation before the approaching EP elections. One useful window into the state of play is provided by a 25 January 2019 blog post by Jens-Henrik Jeppesen and Laura Blanco. There are many problems with the draft Regulation, aggravated by the urgency of the matter.
This blog post addresses only one issue, but one of fundamental importance. With all due respect, it needs to be said that the Emperor has no clothes. The draft EU Regulation on preventing the dissemination of terrorist content online is not about content! The proposed criteria for taking down material from the internet are understandably based on agreed (even if problematic) existing texts concerning terrorist crimes. This creates an image of legality and foreseeability. What is, however, hidden behind that image is that algorithmic or administrative decisions to order immediate removal from the internet of terrorist content cannot, either logically or in practice, be based on criteria that have been written for being determined through the evidence-based adversarial process of a criminal trial.
For the above reason, many seemingly constructive proposals to bring the text of the Regulation even closer to agreed or better-than-agreed legal texts seeking to define terrorist crimes also fail. This is the case for instance concerning a submission by three UN Special Rapporteurs on human rights who have proposed reformulating the criteria in the Regulation on the basis of my own 2011 proposal for a model provision on the crime of incitement to terrorism, made in a report as Special Rapporteur on human rights and counter-terrorism (see good practice no. 8).
My claim about the Emperor not having any clothes is quite simple. Here are the criteria for ‘terrorist content’, as formulated in Article 2 (5) the Council’s agreed text for the Regulation:
(5) 'terrorist content' means material which may contribute to the commission of the intentional acts, as defined in Article 3(1)(a) to (i) of the Directive 2017/541, by:
aa) threatening to commit a terrorist offence;
(a) inciting or advocating, such as [sic] the glorification of terrorist acts, the commission of terrorist offences, thereby causing a danger that such acts be committed;
(b) soliciting persons or a group of persons to commit or to contribute to terrorist offences;
(c) promoting the activities of a terrorist group, in particular by soliciting persons or a group of persons to participate in or support the criminal activities of a terrorist group within the meaning of Article 2(3) of Directive (EU) 2017/541;
[sic] instructing on methods or techniques for the purpose of committing terrorist offences.
About half of the elements of the definition are not related to the content of any verbal or non-verbal communication on the internet. I have placed in italics the clearest cases where the definition is not about content but about intent, context or consequences. Assessing whether a text, image or video may contribute to acts of terrorism (see the chapeau) requires assessment of factors that are not about the content of the matrial. By examining what was posted on the internet one cannot judge whether someone had the intention to glorify a terrorist act, thereby causing an objective danger that someone else commits an act of terrorism (item a). Taking down material based on it being understood as supporting a terrorist group (item c) will result in targeting materials on the basis of their presumed political aims which may be fully legitimate. Only a contextual assessment and evidence presented by the prosecution can establish that a third party not only supported the same political aims (e.g. a people’s right of self-determination) as some terrorist organization, but that it also supported the tactics of terrorism. And many types of information on weapons, hazardous chemicals or explosives are also perfectly legitimate, unless one can show an intent to instruct someone to commit acts of terrorism (see unnumbered last item d which explicitly refers to "purpose").
Many of these elements of draft Article 2 (5) are familiar from the EU counter-terrorism Directive 2017/541 that requires their transpositioning into national penal codes for the purpose of prosecution and trial. In a criminal trial a judge will determine whether the prosecution has presented compelling evidence under each of the subjective and objective elements of the definition of a crime. What actually happened (actus reus) matters, but so does the intent under which the act was committed (mens rea). Therefore copy-pasting even exactly the same wording into a regulation that allows algorithms or human analysts to take down material from the internet cannot be based on the same criteria, as they can only be applied through presenting actual evidence about intent and context.
The draft Regulation constitutes a grave threat to freedom of expression. It could be applied in respect of journalists, non-governmental organisations, political parties, trade unions, indigenous peoples, scholars of history or social sciences, novelists, cartoonists, photographers and filmmakers. Its cross-border application makes it a dreadful tool in the hands of authoritarian regimes or rogue officials.