The International Forum for Responsible Media Blog

Case Law, Strasbourg: Delfi AS v Estonia: Court Strikes Serious Blow to Free Speech Online – Gabrielle Guillemin

DelfiOn 10 October 2103, the European Court of Human Rights handed down judgment in Delfi AS v. Estonia (no.64569/09), a case concerning the liability of a news portal, Delfi AS for third-party comments made on its website.  The Estonian courts found that Delfi AS had editorial control over the comments’ section on its news site and should have prevented clearly unlawful comments from being published, notwithstanding the fact that Delfi had taken down the offensive comments immediately upon being notified of them.

In a unanimous judgment, the First Section of the Strasbourg Court concluded that the domestic courts’ findings were a justified and proportionate restriction on Delfi’s right to freedom of expression.

The Court’s judgment in Delfi is not only a serious blow to freedom of expression online but also displays a profound failure to understand the EU legal framework regulating intermediary liability. In addition, it conveniently ignores relevant international standards in the area of freedom of expression on the Internet. Delfi is a decision that cries out to be overturned on appeal by the Grand Chamber. Failing that, applicants would be well advised to look to the Court of Justice of European Union in Luxembourg to protect their right to freedom of expression online in future, rather than waste their time in Strasbourg.


Delfi is one of the most popular Internet news site in Estonia, publishing up to 300 news articles a day. In January 2006, it published an article online concerning the controversial decision of a ferry company, known as ‘SLK’ to change its routes and the implications of that decision for ice roads between the Estonian mainland and various islands. The article attracted 185 comments, about 20 of which contained insulting and/or threatening language towards the majority shareholder of the ferry company, referred to in the judgment only as “L”. On 9 March 2006, L.’s lawyers requested Delfi to remove the offensive comments and claimed 500,000 kroons (approx. 32,000 euros) in damages. Delfi complied with the request on the same day but refused the claim for damages.

In April 2006, L. brought civil proceedings against the applicant company. Although unsuccessful at first instance, he obtained judgment against Delfi in June 2008 and was awarded 5,000 kroons (approx. 320 euros). Judgment was confirmed by the Tallinn Court of Appeal and ultimately by the Supreme Court in June 2009. The Estonian courts found that the comments at issue were defamatory and/or insulting and as such could not be said to be protected by the right to freedom of expression.  Importantly, they concluded that Delfi should be considered as a publisher of the comments because it had control over the comment section of the website. In particular, it had adopted house rules in relation to the comment section and removed comments that were found to be in breach of those rules. For this reason, the domestic courts rejected Delfi’s argument that it was immune from liability under the EU Directive 2000/31/EC as it could be not be said that Delfi was a purely technical, and therefore, passive intermediary. Another significant element in the domestic courts’ reasoning was that Delfi drew economic benefits from the comments that were made in the comment section of its website. Finally, the domestic courts considered that Delfi should have prevented clearly unlawful comments from being published and after their publication, should have taken them down of its own initiative.

Chamber Judgment

In a unanimous judgment, the Chamber of the First Section concluded that there had been no violation of Article 10 ECHR. The Court broadly endorsed the reasoning of the domestic courts.

First, the Court clarified that it was not its role to substitute its own interpretation of domestic legislation to that of the domestic courts or address the position under EU law.

It was not disputed that the comments posted by readers were defamatory or even unlawful. Nor was it contested that the applicant company promptly removed the ‘infringing’ comments. The key issue was whether Delfi’s civil liability for the defamatory comments was a disproportionate interference with its right to freedom of expression.

In addressing this question, the Court first examined the context of the comments. Although the Court acknowledged that the news articles itself was balanced and addressed a matter of public interest, it considered that Delfi “could have realised that it might cause negative reactions against the shipping company and its managers” and there was “a higher-than-average risk that the negative comments could go beyond the boundaries of acceptable criticism and reach the level of gratuitous insult or hate speech.” Accordingly, the Court concluded that Delfi should have exercised especial caution to avoid liability.

Next, the Court examined the steps taken by the applicant company to deal with readers’ comments. In particular, the Court noted that Delfi had put in place a notice-and-takedown system and an automatic filter based on certain key ‘vulgar’ words. However, the Court concluded that the filter in particular was “insufficient for preventing harm being cause to third parties’.  Although the notice-and-takedown system was easy to use – it did not require anything more than clicking on a reporting button – and the comments had been removed immediately upon notice, the comments had been accessible to the public for six weeks. The Court considered that the applicant company “was in a position to know about an article to be published, to predict the nature of the possible comments prompted by it and, above all, to take technical or manual measures to prevent defamatory statements from being made public”. Since the actual writers of comments could not modify or delete their comments once posted on the Delfi news portal, Delfi effectively exercised sole control over the publication of comments even if it did not exercise such control to its full extent.

The Court went on to note that Delfi had been given leeway by the domestic courts as to how it should ensure the protection of third-parties rights. Indeed, the domestic courts had not prescribed prior-registration of users or pre-moderation of comments. Rather, they had imposed a fine of 320 euros, which given Delfi’s position as a professional operator of one of the largest Internet news portals in Estonia was ‘by no means’ disproportionate.

Finally, the Court considered that it would be disproportionate to put the onus on complainants to identify the authors of anonymous comments. By allowing comments to be made by non-registered users, Delfi had assumed a certain responsibility for them. The Court further noted that “the spread of the Internet and the possibility – or for some purposes the danger – that information once made public will remain public and circulate forever, calls for caution”. In the Court’s view, it was a daunting task to identify and remove defamatory comments at the best of times, including for the applicant. It would be even more onerous for a potentially injured person, “who would be less likely to possess resources for continual monitoring of the Internet”.

The Court concluded that there had been no violation of Article 10 of the Convention.


The decision of the First Section sets a deeply worrying precedent for freedom of expression in several respects. It also displays a worrying lack of understanding of the issues surrounding intermediary liability and the way in which the Internet works. All the more disturbing is that the Court’s decision in this case was unanimous (although tellingly several judges sitting in the Chamber came from non-EU countries, namely Azerbaijan, Macedonia and Russia, and an EU-newcomer, Croatia).

While the Court did not seek to address the position under EU law, a basic understanding of the provisions dealing with intermediary liability under the E-Commerce Directive (‘ECD’) was fundamental to a proper examination of the case, including the notice-and-takedown system and the use of filters “to prevent violations of third party rights”. However, the Court utterly failed to appreciate that the purpose of the provisions concerning hosting liability (Article 14 ECD) is to give an incentive to platforms such as Delfi to remove content upon notice of its illegality and that, in exchange, those platforms are given immunity from liability. If they fail to take action upon notice, they could then face liability for the content at issue. In the present case, Delfi had every reason to believe that it fell within the scope of protection under the ECD since it had removed the material on the same day that it had received the complaint. That should have been the end of the matter. However, both the Court and the domestic courts considered that this was not going far enough. What Delfi ought to have done was to prevent defamatory and other ‘clearly unlawful’ comments from being made public, disregarding Article 15 ECD (set out in para. 41 of the Court’s judgment) which prohibits Member States from imposing monitoring obligations on information society services, including ‘actively to seek facts or circumstances indicating illegal activity’.

Not only did the Court’s judgment fail to grasp the EU framework governing intermediary liability but it also conveniently ignored relevant international standards developed by the UN Special Rapporteur on Freedom of Expression in this area in his thematic report on the Internet (A/HRC/17/27). In particular, the UN Special Rapporteur recommended that “censorship measures should never be delegated to private entities, and that no one should be held liable for content on the Internet of which they are not the author.” The UN Special Rapporteur was very critical of notice-and-takedown procedure, notably the fact that material is removed without a judicial determination of the question whether the content at issue was indeed unlawful. But the Court did not concern itself with such matters. On the contrary, the Court gave a nod of approval to ‘reporting button’ – type of procedures, which do not even require the slightest explanation as to why the content at issue might be defamatory or, better still, unlawful. That the level of detail of takedown notices has been the subject of extensive criticisms and debate across the EU, for instance in the context of the Defamation Act 2013 in the UK, was again entirely ignored. Nevermind also that internet intermediaries are not best placed to make these types of determinations, let alone a filter.

Worse still, the Court concluded that the Internet filter used by Delfi to delete vulgar words was ‘insufficient’ to protect the rights of third parties and that the notice-and-takedown procedure used by Delfi had allowed the offensive comments to remain accessible for 6 weeks, implying that Delfi should have at least removed the comments as soon as they were posted. That the type of technical measures required to ‘prevent’ third-party violations favoured by the Court would force Internet intermediaries to pre-moderate content or inevitably block access to lawful material was however entirely lost on the Court – and this, despite a recent ruling of the Court of Justice of the European Union confirming this in the SABAM v Netlog case (the Court itself mentioned Scarlet Extended at para. 45 which made a similar point in relation to ISPs).  In fact, the Court took pains to indicate that an important consideration in its judgment was that the domestic courts had not ordered an injunction requiring pre-registration of users or pre-moderation by the applicant. In other words, the Court entirely failed to follow its own findings to their logical conclusion. Moreover, the Court conveniently overlooked that any damage suffered as a result of the defamatory comments being accessible for 6 weeks was the direct result of the complainant’s own failure to notify the material to Delfi despite the availability of a swift and easy-to-use reporting procedure.

More generally, the Court seems to have naively assumed that every injured party acts in good faith, the reported content is always in fact unlawful and Internet filters are the silver bullet to deal with all the different types of illegal content that circulate on the Internet. However, as has been pointed out time and again, the Internet fairy and its magic wand is a myth.

The Court also made a number of worrying statements, including the suggestion that Internet news portals should realise that their articles might “cause negative reactions”, some of which might go beyond the bounds of acceptable criticism and that therefore they should be prepared to take the necessary measures to avoid liability. For anyone familiar with the way in comments online operate on news sites, this is laughable. The vast majority of public interest news will almost by definition stir debate and attract comments of all kinds, including offensive ones. While it may be appropriate for those sites to remove insulting comments upon notice in accordance with their house rules, what the Court is suggesting is that internet news portal have knowledge of illegal content on their platforms ‘by default’ and should take steps to prevent their publication or be prepared to face the consequences. Short of all out private censorship, the upshot of the Court’s judgment is that news portals should close their comment section to avoid liability. This would both deprive Internet users of a valuable way of engaging in matters of public debate and news sites from generating revenue at a time when their business model is seriously threatened.

Following the Court’s logic, its conclusion that platforms such as Delfi should bear the responsibility for anonymous comments may not be as bad as it seems in terms of its practical effect. If platforms are made responsible for any potentially illegal comment regardless of any notice-and-takedown procedure that may be in place, then a requirement to pre-register with a platform would do nothing to help intermediaries avoid liability. Of course, as noted above, the consequences of the Court’s logic are far worse since they are likely to encourage news portals to close down their comment section. At any rate, the Court displays an unsurprisingly conservative disposition towards the Internet and the “dangers” of information remaining publicly available “forever” on the internet. If the Court’s decision is allowed to stand, one can only fear the outcome of an application brought by Mr Mosley against the likes of Google.

Ironically, the Court seems to have been concerned to protect individual’s rights online against powerful companies. While this may have been well-intentioned, the Court fundamentally failed to understand the role of Internet intermediaries as the gateways to the exercise of free expression. The Court also made much of the fact that the applicant company was a commercial operator rather than an individual blogger moderating his or her blog, for instance. However, the Court failed to explain the nexus between drawing advertising revenue from readers’ comments and taking responsibility for that content.

Finally, while it may be true that the award of damages imposed on Delfi was relatively small, the point is that it should never have been held civilly liable for the offensive comments when it had removed them immediately upon notice. To hold otherwise was to defeat the purpose of the defence stemming from the ECD, with a highly likely chilling effect on the free flow of information online. As such, it was a disproportionate interference with Delfi’s right to freedom of expression. While it may not have been within Strasbourg’s competence to rule on the legal position under EU law, its role surely was to tell the domestic courts that they got it wrong. Instead, the Court has delivered a highly short-sighted and damaging judgment for freedom of expression online.


For anyone familiar with the EU legal framework governing intermediary liability, this ruling will come as a surprise. It is also almost certainly likely to create even greater legal uncertainty in this area. It is worth remembering that in its first judgment in this case, the county court had initially found in favour of the applicant company. As Lord Justice Ward once said, “this goes to prove what every good old-fashioned county court judge knows:  the higher you go, the less the essential oxygen of common sense is available to you”. Let’s hope that the Grand Chamber will prove him wrong. Otherwise, applicants in these types of cases would do well to go to Luxembourg rather than Strasbourg to seek protection for their right to freedom of expression online.

Gabrielle Guillemin is legal officer at ARTICLE 19, an international free speech organisation. She formerly worked as a lawyer at the European Court of Human Rights. For more information on the work of ARTICLE 19 on intermediary liability, see Internet Intermediaries: Dilemma of Liability (2013)


  1. Prof Gavin Phillipson

    Excellent analysis of what sounds like a very poor decision. One for the UK courts to decline to follow I think.

  2. davidhencke

    Extraordinary ruling. I have written a blog on my own site drawing attention to it and linking to Inforrm so that bloggers can be made more aware of this

  3. Evan Harris

    I agree with this analysis almost entirely. It would appear to drive a coach and horses through the protection for secondary publishers provided by the E-Commerce Directive and the sensible EU and domestic case-law emanating from that. As the post says, the E-Commerce Directive provides that secondary publishers (for want of a better phrase) are only potentially liable if they have actual knowledge of UNLAWFUL content and the ability to remove it.

    But I think that Gabrielle is wrong to say “If the Court’s decision is allowed to stand, one can only fear the outcome of an application brought by Mr Mosley against the likes of Google”. I am no expert on this, and make no judgement on the merits, but I think Max Mosley’s case is different surely. It is predicated on the fact that the content has been found in court to be unlawful and the hosts have actual knowledge of this and can remove it. Unlike this case where liability is being posited on the basis of allegedly defamatory (not the same as unlawful) material, and prior even to actual knowledge of its defamatory nature.

    Declaration: I am a member of the Libel Reform Campaign (which campaigned for stronger defences for secondary publishers in English law), and Associate Director of Hacked Off.

  4. Victoria McEvedy

    Agree that it is worrying but the fact the defendant was a news organisation in the real world seems to have greatly influenced the decision. When the Ecommerce Directive was first introduced there was doubt about whether newspaper sites should be entitled to the defence –on the basis that they should assume responsibility for all publications including readers’ comments –just as they had always done when publishing letters to the editor. This underlies the restricted nature of the secondary responsibility defence in the 1996 Defamation Act with its exclusion of commercial publishers from secondary status. This was resolved here as the courts did not force the newspapers to use the narrower Defamation Act defence –see Imran Karim v Newsquest Media cited with approval in Kaschke v Gray. In all, they seem to have regarded the case as wholly exceptional in the amount and nature of the abuse —which is telling as it would not be so here. English law would also regard most quoted postings in the case as mere Vulgar Abuse. Agree –it’s not a well-reasoned judgment at all. Interestingly, our new defence in section 5 of the Defamation Act 13 (website operators) –also avoids providing immunity for intermediaries who continue to publish material where authors are not prepared to provide their identity –but they have to be needing a defence (be a publisher) in the first place under domestic law.

  5. lyndaphillips2

    They can’t take my words, they’re mine… I have nothing really left to lose; so they can sue me if they want to waste their money…

  6. Andrius

    I’m trying to persuade myself that the decision of the ECHR is on the article 10 only and should be analysed only in the light of the EHRC. The court ruled that the *Delfi’s right* to freedom of expression was not infringed while the questions of Delfi’s liability was not touched. Thus, I believe the decision should not make a precedent in the ECJ or European national courts. However, I am looking forward to the first EU member court which forwards this question to the ECJ.

Leave a Reply

© 2023 Inforrm's Blog

Theme by Anders NorénUp ↑

%d bloggers like this: