New documents reveal the truth behind the Hate Speech Code

EDRi obtained documents revealing different drafts of the Code of Conduct against Hate Speech and the correspondence exchanged between the European Commission and the four big companies that concluded the agreement on 31 May 2016: Facebook, Google, Twitter and Microsoft. These documents complement an interesting response sent by Commissioner Jourová to a letter from the Center for Democracy & Technology (CDT) on 21 June 2016.

These are our main findings:

1. The Commission doesn’t believe in multistakeholderism

While the European Commission claims to support a multistakeholder internet governance model, it decided it did not want to include other parts of industry, civil society, academics or any other relevant stakeholder in the elaboration of the Code of Conduct against Hate Speech. Indeed, it only shared the Code with the 28 Member States a few days before its launch, which led to public authorities not having a possibility to suggest any changes to the text. Although EDRi and Access Now knocked on the Commission’s door several times, the European Commission only accepted to hold bilateral meetings, refusing to give us access to the negotiations and the drafts. As a result, on 28 April we asked the Commission to send us all the documents and conversations relevant for the adoption of the Code.

2. If you obey the law, we still won’t leave you alone

In the letter to CDT, Commissioner Jourová argues that the definition of hate speech is clear. She seems to believe that it agreed with companies to comply with the law (the Framework Decision on on Racism and Xenophobia).

The Commission claims that the law has “addressed the fragmentation” of what constitutes illegal hate speech among Member States. However, a recent project financed by the European Commission warned about the “huge disparities” on what constitutes illegal hate speech among Member States. A recent study of the European Parliament confirms this. The Commissioner tries to argue that the Framework Decision draws a “careful line” between illegal hate speech and freedom of expression. However, as EDRi member Article 19 demonstrated, this is sadly not true; international law standards are not respected.

In addition, documents obtained by EDRi show that Facebook, Google and Twitter repeatedly contended that all referrals would be assessed by their terms and conditions, not by the law. They would only review hate speech reports against the law when a potential violation is not identified as a terms of service violation. In short, companies did not commit to follow the criteria set out in the law. They committed to use their own criteria in deciding what is or is not hate speech. Contrary to the Commission’s assurances, the Code can lead to “undue restrictions” to freedom of expression and “excessive take downs also of legal content” because restrictions imposed by terms of service are, quasi by definition, much broader than the law.

According to the Commission, if an online company restricts your freedom of expression – even if they do so as a result of government pressure – you are not protected by the Charter of Fundamental Rights of the European Union. You would only be covered if the company was implementing a specific legal obligation. Does the Code impose a legal obligation on companies? No. What’s more, the terms and conditions of companies will not be reviewed by the Commission for this purpose. Despite the Commission’s assurances, there are no protections whatsoever for fundamental rights in this code. The Commission is, at least, consistent, as the same rights-destructive methodology can also be found in the leaked copyright Directive, the draft Audio-Visual Media Services Directive, the Europol Regulation, the “follow the money” copyright enforcement projects…

3. Root of the problem unsolved: there’s nothing new under the sun

The Commission claims that since hate speech is moving online, the enforcement of the law must be complemented by companies’ actions to be “quick”. While the Commission’s goal was for the companies to commit to remove clear cases of hate speech within 24 hours, the companies only committed to review (not remove) illegal hate speech (not term of service violations) within 24 hours.

These “complementary efforts” are however not new. The EDRi-gram reported about the flaws of privatised enforcement already over a decade ago. Sadly, the discussions keep running in circles: a public policy problem is identified; public authorities put pressure on companies to solve it (even after the adoption of the Code, countries like Germany keep putting pressure on social media companies to do the job of a public authority); the content is the target, not the author(s) of the content; Member states do not investigate or prosecute; the Commission does not take responsibility; the root of the problem is not solved. Only if the problem is the need to be seen to be doing “something”, the problem is solved.

As the Commission confirmed in its letter to CDT, “the Code does not deal with matters pertaining to criminal law proceedings against the authors of hate speech … or notification of illegal content to public authorities.” The actual problem becomes irrelevant in the Code.

The Code of Conduct is divided into two parts, a descriptive section and an operative section. The Commission “won” the negotiation with industry on what should be in the descriptive part. There are lots of nice words about concepts and about fighting “illegal” content. However, the Commission lost and the companies won on the operative part. In that section, references to “illegal” content suddenly become sparse, the companies undertake to review first on the basis of their terms of service and only, where necessary (i.e. never, as illegal content is banned by terms of service) on the basis of the law. The descriptive section does not describe what is in the agreement, it describes what the Commission wished for. The operative section describes the loss of those wishes. The Commission won the descriptive section, the companies won the operative section while the rule of law, free speech, internet users (especially victims of hate speech) lost in both sections.

EDRi’s Freedom of information request to DG Justice on the Code of Conduct against Hate Speech and responses from the Commission (28.04.2016 and 28.07.2016, respectively)
https://www.asktheeu.org/en/request/code_of_conduct_against_hate_spe

EDRi: Guide to the Code of Conduct on Hate Speech (03.06.2016)
https://edri.org/guide-code-conduct-hate-speech/

Commissioner Jourova’s response letter on the Code of conduct on illegal online hate speech (21.06.2016)
https://cdt.org/files/2016/09/Commissioner-Jourova-to-Mr-Jeppesen.pdf

Mandola Project: Intermediate report. Definition of illegal hatred and implications (20.07.2016)
http://mandola-project.eu/m/filer_public/7b/8f/7b8f3f88-2270-47ed-8791-8fbfb320b755/mandola-d21.pdf

Germany wants Facebook to take initiative in fight against online hate (29.08.2016)
http://www.reuters.com/article/us-germany-facebook-idUSKCN1141S6

European Parliament Study. The European legal framework on hate speech, blashemy and its interaction with freedom of expression (04.09.2015)
http://www.europarl.europa.eu/RegData/etudes/STUD/2015/536460/IPOL_STU(2015)536460_EN.pdf

Press release: Counter-Extremism Bill, National Security Council Meeting (13.05.2015)
https://www.gov.uk/government/news/counter-extremism-bill-national-security-council-meeting

(Contribution by Maryant Fernández Pérez, EDRi)