Document Actions

No section

EU Copyright Liability for Internet Linking

  1. Pekka Savola


This article analyzes linking on the internet from the perspective of the EU copyright concept communication to the public in the light of recent developments in late 2016 and early 2017, especially the Court of Justice of the European Union judgments GS Media, Filmspeler, and Ziggo. The article highlights the doctrinal approach on communication to the public and de facto harmonization of certain aspects of classic indirect liability. The article analyzes open issues relating to linking to illegal publications, specifically the uncertainty relating to the extent of liability for the contents of the linked page, the scope of pursuing profit when linking, and the linker’s duties of care. The article also discusses two aspects related to linking to legal publications; that is, the conceptualization of the “new public” requirement as a regulated implied license and the conditions of restricting access and circumventing restrictions.


1. Introduction*

Article 3 of the Information Society Directive (2001/29/EC) provides, based on Article 8 of the World Intellectual Property Organization (WIPO) Copyright Treaty (WCT) of 1996, that “Member States shall provide authors with the exclusive right to authorise or prohibit any communication to the public of their works, by wire or wireless means, including the making available to the public of their works in such a way that members of the public may access them from a place and at a time individually chosen by them.”  [1]

The first part concerns the original communication to the public, the second part a “sub-right”, so-called making available. The latter refers to providing access for on-demand transmission, and it is more extensive in this context because it applies irrespective of transmission occurring. During the WCT treaty preparations, making available was referred to “offering” and “providing access to a work”, and it was coined to prevent uploading and making protected files accessible online.  [2] At the time, there was no consideration of making available covering hyperlinking to a work already legally or illegally available elsewhere.  [3] Linking as such has also not been subject to EU legislation, and it would be up to the judiciary to deal with any upcoming issues.

This article analyzes linking on the internet from the perspective of the EU copyright concept “communication to the public” in the light of the case law of the Court of Justice of the European Union (CJEU), and the GS Media judgment in particular.  [4] The CJEU’s approach is taken as a given and therefore the focus is on interpretation and mapping (and even filling) the blanks. In the interest of brevity, discussion of earlier judgments and communication to the public in general is kept to a minimum.  [5] The unharmonized moral rights and other copyright aspects, such as the reproduction right, as well as domestic remedies on non-copyright grounds, are likewise out of scope.  [6]

The rest of this article is structured as follows. Part 2 summarizes the key points and uncontentious findings of internet linking case law up to and including the GS Media judgment.  [7] Part 3 highlights the two most important and general doctrinal aspects raised by the recent judgments: the conceptualization of communication to the public and its expansion towards also covering indirect liability for the acts of others. Part 4 discusses fundamental practical open issues, specifically the uncertainty of the extent of liability for the contents of the linked page, the scope of pursuing profit when linking, the linker’s duties of care, as well as the conceptualization of the “new public” requirement as a regulated implied license and the conditions of restricting access and circumventing restrictions. Part 5 concludes this article.

2. Summary as of GS Media

The GS Media case concerned the liability of an online website for the articles that included hyperlinks to a file sharing website where unpublished photographs had been uploaded without the consent of the right holder. Advocate General Wathelet suggested redefining the course set in the earlier case law so that linking would not be an act of communication, and even if it were, it would not be a communication to a public.  [8] The CJEU was not willing to accept either proposal. Instead, it opted to invent a new condition to be imposed on top of the latter alternative.

The Court recalled several complementary and independent criteria, which may in different situations be applied individually and in their interaction. The first and most important of them was the indispensable role played by the user and the deliberate nature of its intervention.  [9] The profit-making nature of a communication was also deemed relevant.  [10]

Further, the Court explained that the previous internet linking cases, Svensson and Bestwater, were meant to apply only in cases where the works had already been made available with authorization, and in those cases the act of communication was not made to a new public.  [11] In essence, Svensson had allowed linking to a freely accessible protected work published with the consent of the author, and Bestwater confirmed that all forms of linking are treated equally.  [12] Indeed, “as soon as and as long as that work is freely available on the website to which the hyperlink allows access”, the copyright holders have consented to such a communication, i.e., linking.  [13] In contrast, it could not be inferred that linking to non-consented publications would not be communication to the public.  [14]

The Court observed that the internet is of particular importance, taking into account that automatically considering all linking to non-consented publications as communication to the public would have highly restrictive consequences on the freedom of expression and of information. Further, the Court contemplated the difficulty of ascertaining the consent and practical difficulties in verifying the legality.  [15]

This conundrum led the Court towards an individualized assessment to define an entirely new set of conditions. In essence, the liability for linking to works published without authorization depends on a reasonable knowledge standard. A person who does not pursue profit when posting a link (non-commercial linker), is not as a general rule, expected to be aware of the lack of authorization. When posting links is carried out for profit (commercial linker), the expectation is to carry out necessary checks to ensure the work is not illegally published on the website to which those hyperlinks lead, which amounts to a rebuttable presumption of knowledge. The knowledge standard “knew or ought to have known” can also be established with other means, for example, an explicit notification of illegality. That is, non-commercial linkers may be shown to have had actual knowledge, and commercial linkers can exonerate their liability by rebutting the construed knowledge.  [16]

Moreover, the links that allow circumventing the restrictions taken by the site to only subscribers constitute communication to the public.  [17] On the other hand, the Court also noted that there would not be a new public if the works have been made freely available on another (apparently actually any other) website with the consent of the right holder.  [18] These conditions which allowed the right holders to take action against either linkers or original publications were deemed to provide the required high level of protection for authors.  [19]

The judgment also essentially stated that upon a notification of illegality, the linkers cannot rely upon the copyright exceptions of Article 5(3) of the Information Society Directive. [20] The opening paragraphs included Article 5(3)(c), on reporting on current events by the press, but no specific copyright exception was further discussed. However, the lawfulness of the source is not required by that specific exception or EU copyright law in general unless derived from Article 5(5), and in consequence the basis and impact of this statement is left vague. [21]

3. Shaking the Doctrinal Foundations

3.1. The liability and foundations of communication to the public

In the preceding Svensson doctrine, all linking acts are in principle making available and therefore acts of communication. These acts of communication almost always have an indeterminate number of potential recipients and are therefore also communication to a public.  [22] These do not necessarily attract a new public, however, and then there would not be communication to the public.  [23] In Svensson, the focus was on the new public, rather than qualifying the linker’s “intervention” and considering why the linking is an act of communication in the first place.  [24]

In GS Media, the issue is characterized as whether the linker intervenes in full knowledge (construed or actual) of the consequences of his conduct in order to give access to an illegal publication. If such knowledge is missing, there is no communication to the public.  [25] While the AG advocated a position that linking must be indispensable – as in vital - to access the works the CJEU only referred to the indispensable role in the context of previous case law and only used “deliberate intervention” in the context of circumventing access restrictions.  [26] Essentially the CJEU appears to have softened the requirement especially when compared to its literal meaning.  [27]

In case of linking to illegal publications, GS Media did not take an implicit or explicit doctrinal stance on the partitioning into different “components” of communication to the public, except by excluding the new public condition. Thus, one was left to wonder whether the deliberate intervention test would be an alternative only to the new public requirement, the previous communication to the public doctrine entirely, or even something else.  [28]

AG Opinions in Filmspeler and Ziggo examined and held that there is a new public, even though in both cases linking concerned unconsented publications.  [29]

In Filmspeler, the CJEU held that the sale of the multimedia player was considered an act of communication, it targeted a public, it was made in full knowledge of the fact of hyperlinks giving access to works published illegally on the internet, and was done with a view of making a profit. The CJEU specifically discussed but distinguished the case from ones finding a new public.  [30]

In contrast, in Ziggo the CJEU held that the making available and management of a file sharing platform was an act of communication, protected works were communicated to a public, the platform provided access to works published without the consent of the right holders, and there was communication to a new public.  [31] Further, the making available and management of the online sharing platform was carried out with the purpose of obtaining profit.  [32]

In the light of Filmspeler, it seemed clear that for all kinds of linking, first the conditions on “act of communication” and “a public” need to be examined. Then linking to consented publications must fulfill the “new public” requirement, and linking to unconsented publications the deliberate intervention and construed or actual knowledge requirements. The most plausible explanation would seem to be to disregard the new public elements in Ziggo as an unnecessary doctrinal misstep.  [33] Alternatively, linking to illegal sources (as was the case in GS Media and indirectly in Filmspeler) might be evaluated based on a different doctrine than communication to the public in general (which may or may not have been the case in Ziggo).

The doctrinal consequence of GS Media and Filmspeler seems to be that the “volition” or “act” (understood widely to encompass the whole communication to the public) disappears if the knowledge standard is not met. The implications remain to be explored. This is a somewhat unexpected doctrinal situation, because typically the lack of knowledge may result in the person not being liable, and not nullify the illegality of the deed.  [34]

A doctrinal alternative might have been that the act of linking remains unlawful, but the lack of negligence or required intention may result in the lack of liability, as with criminal law and torts in general. However, in the EU framework, there is only limited harmonization of liability and damages.  [35] Adopting this alternative would have required overriding domestic liability standards, which typically apply strict liability for direct copyright infringement, by enforcing intention-based liability. Taking this and the structure of communication to the public right into account, the CJEU did not have much room to manoeuvre to take into account doctrinal aspects. On the other hand, the third option would have been to refuse holding infringement of communication to the public, falling back to the various domestic secondary liability doctrines, or awaiting the EU legislator’s future initiatives.  [36]

One of the most fundamental consequences is the evaluation of liability in this context. While the CJEU does not classify communication to the public as direct and indirect liability, the knowledge requirements make this more akin to indirect liability. Even though, from the domestic perspective, liability may still be strict, fewer acts would be considered infringing in the first place, leading essentially to a similar result as with intention-based liability regimes.  [37]

3.2. De facto harmonization of secondary liability

“Intervention, without which, … in full knowledge of the consequences” that originated in SGAE  [38] is in fact secondary liability of a sort: a deliberate and conscious act that leads with likelihood to an infringement by someone else.

EU law does not harmonize secondary liability and there are varying domestic approaches.  [39] However, the AG opined in Ziggo that the solutions must be sought in EU law, rather than left to the devices of domestic legal systems (and therefore their secondary liability regimes).  [40] After the sale of a specific kind of multimedia player had been deemed communication to the public in Filmspeler, there was little doubt that there would also be communication to the public in Ziggo. Indeed, the CJEU was willing to find solutions in EU law, even by extension and expansion, rather than leave the issues to be addressed by domestic law and the EU legislator.

Consequently, in practice de facto harmonization of secondary liability is already under way in these specific conditions.  [41] The CJEU further defined the “knowledge” or culpability-related aspects of secondary liability in GS Media, Filmspeler, and Ziggo. Further clarifications are likely to follow and domestic judgments will probably continue to apply some of the domestic secondary liability principles.  [42] The CJEU also took a stance on the indispensable role, that is, the causal element typically examined in secondary liability.

4. Selected Open Issues

4.1. The extent of liability for the linked page

4.1.1. Introduction

Web pages are technically text files, possibly literary works on their own, which almost always include clickable links to other web pages and Hypertext Markup Language (HTML) instructions for users’ web browsers to load images or other elements from the same or other websites (inline linking).

Is the linker also responsible for linking (with construed or actual knowledge) to websites which include inline or clickable links to infringing material? That is, is the linker also responsible for elements whose loading was caused by the author of the linked page or, respectively, the possibility of users to access other webpages through clickable links? Does the ambit of the commercial linkers’ reasonable checks include verifying these elements, or can the non-commercial linkers’ knowledge of these be established?

These questions can be examined using two scenarios depicted in Figure 1. The bottom one, “Link Page Example”, concerns linking to a page including clickable links to illegal content elsewhere. The top one, “Blog Example”, concerns linking to a page also including an inline-linked image from an unlawful source as well as a link to the link page.

Figure 1: examples for discussing the scope of liability when linking to a web page

The factual background in GS Media was that the first link had been set to a website from whereby clicking on another link the photos at issue could be found.  [43] However, later in the judgment this is described differently as “provid[ing] the hyperlinks to the files containing the photos at issue”.  [44] The second hyperlink provided access to a website “where one or more of the relevant photographs could be viewed”.  [45] The third article “contained a hyperlink to the photos at issue” and “forum users of that (the referring) website then posted new links to other websites where the photos at issue could be viewed”.  [46] In the latter two cases, the wording suggests that the photographs may have been directly available on the target website without further clicking, but the phrasing leaves this ambiguous. Given the factually inaccurate and inconsistent descriptions of the first linking case one is left to wonder whether the CJEU grasped this distinction – or understood it but deemed it irrelevant and hence seemed to treat these in the same manner.

Nonetheless, the referred question concerned a hyperlink to “a website on which the work has been made available”. The CJEU interpreted the question to mean, in essence, “hyperlink to protected works, freely available on another website”. This distinction could be extremely important depending on whether “website” is deemed to mean only the directly linked element, everything on the linked page, or even everything available on the target website through clickable links. The CJEU seems to mainly discuss the case when the hyperlink directly refers to protected works, but with a wider interpretation of “website” it is possible to also consider the target page or website as a whole.  [47]

4.1.2. Possible interpretations

For illustrative purposes, let us consider four different interpretations and scopes of infringement. Other variations are also possible and therefore the list is not exhaustive. Unfortunately, the inconsistencies and legal and policy implications make it very hard to both interpret and predict future application, and especially CJEU interpretation, of GS Media criteria. Nonetheless, some options are considered.

With a narrow interpretation, the linker could only be liable for the directly linked element itself. In the case of the Blog Example, this would be the HTML text of the blog, but not the images or other elements of that site.

With a medium interpretation, the linker could be liable for the directly available contents on the linked page. In the case of the Blog Example, this would also include the inline-linked images from the same or other websites as they visually appear as part of the linked page. However, the liability would not cover hyperlinks requiring further user intervention. Therefore, there would not be liability in case of the Link Page Example.

With a broad interpretation, the linker could also be liable for everything available on the linked page, including clickable links on the same website but not links to other websites. This would also cover liability in case of the scenarios in GS Media.

With the broadest interpretation, the linker could also be liable for clickable links to other websites. This would also cover the Link Page Example and the Blog Example when it comes to the external hyperlink.

4.1.3. Discussion

Technically and for clarity of the scope of liability, the narrow interpretation would be preferable. However, this interpretation would not even cover the most expansive interpretation of the factual background of GS Media and the CJEU likely did not intend such a narrow scope of protection. This approach must therefore be rejected.

Likewise, the broadest interpretation offered would incur very extensive liability for the original linker. Even if such interpretation might be deemed useful from a legal policy perspective (consider the Link Page Example), the implications for the freedom of expression would be drastic because the linker would need to verify that the linked page does not include any links to infringing sites. Commercial linking to a legal website “A” would result in infringement, if “A” linked to any non-authorized website “B”. Such chain reactions would seem disastrous when considering reasonable duties of care. With this interpretation, linking to the GS Media articles themselves might also have been infringing. This approach must also be rejected, unless a very low threshold of rebutting knowledge of illegality is accepted. If some protection is deemed necessary, one could in some cases require verification of only those external hyperlinks that directly reference copyright works.

GS Media included a phrase “illegally published on the website to which those hyperlinks lead” in the context of verification.  [48] Such wording would seem to explicitly direct the examination only to the linked website (whatever that is considered to include). To cover the factual scenario of GS Media, this would have to also include hyperlinks within the target website but not hyperlinks to external websites. It is not clear if inline linking would likewise be covered.

If we were to ignore the actual factual scenario of GS Media, taking into account that the preliminary question was rephrased to concern directly linking to copyright works, it would be preferable if clickable links that refer to the same website would also be rejected. From the perspective of knowledge and deliberate intervention, additional clicking requiring the volition of the web user could be significant.

If the linker would attract liability for inline linking conducted by the target website (Blog Example), the linker would need to conduct visual inspect of all the contents on the web page. This is also somewhat problematic, but not necessarily disastrous. Again, inline linking from a different website could be distinguished.

4.2. The scope of pursuing profit when linking

The CJEU did not elaborate what would qualify as pursuing profit when linking. The spectrum of financial gain is broad and may include direct and/or indirect profits to varying degrees.  [49] The main distinction is whether the website where the link is provided is run with profit-making purpose or a website operator intends to profit from posting that specific link.  [50] The former option has been exclusive in the literature and has been considered more in line with earlier CJEU case law, and this is how domestic courts in Sweden and Germany have applied the GS Media judgment.  [51]

Let us also consider an alternative. The wording in GS Media  [52] requires that posting hyperlinks must be done for profit. However, it is notable that the CJEU did not phrase this as setting links when pursuing an economic activity, which had already been established in Papasavvas to also cover remuneration through advertising.  [53] If the latter option were to be adopted, this could imply that there may be (otherwise) commercial activities in which setting links is incidental or irrelevant to obtaining profit. If so, at least a weak link of causation would be needed.

That is, the judgment could also be read so that the focus would be in the act of setting links, although not necessarily that specific link. The GS Media case concerned a typical “clickbaiting” headline to attract the users to the articles by advertising the availability of photographs through hyperlinks.  [54] The website was a top 10 news site in the Netherlands and it was undisputed that the website operated for profit.  [55] Therefore, at least setting links on such a popular site to attract users with the hope of getting advertising income qualifies as commercial linking.  [56] Paid subscriptions are clearly not required. A site that is operating as part and related to other commercial activities likely also qualifies automatically. Therefore, it is not obvious that the CJEU necessarily intended the broadest possible interpretation of commercial activity.

In any case, an important question would be whether setting links on any blog including advertising could qualify as pursuing profit.  [57] Typically the answer in the literature and domestic case law appears to be in the affirmative. As before, it seems likely that the standard may have been intended to be set somewhat higher, for example in actively seeking and obtaining financial gain. For example, when linking is casual and not the main object of the site, and there are only a few advertisements which barely pay off the expenses of keeping the site or blog online, this might not necessarily be considered commercial linking.  [58]

Even if the issue of pursuing profit requires either an affirmative or negative answer,  [59] its implications should still be more diverse. Specifically, it would be logical that the weaker the pursuit for profit and the scope of the associated economic activity is, the less would be required as a duty of care (more on this in Part 3.3). In eBay, the standard of a diligent economic operator was established.  [60] It is not very far-fetched to argue that the diligence requirements may differ based on the type of activity and that different kind of operators may have different degrees of due diligence.  [61] Therefore, even with a broad interpretation of commercial activity, the problems could be somewhat mitigated by requiring only a very modest rebuttal (“I checked the website and it seemed to be OK”) especially if there is nothing obviously illegal on the linked web page. On the other hand, for example, very significant economic activity, very clear causative element between setting a link  [62] and pursuing profit or possibly in some cases of journalism, a higher of standard of care could be required.

4.3. The linker’s duties of care

4.3.1. Introduction

The linker’s prima facie infringement is based on construed (commercial linking) or actual (non-commercial linking) knowledge standard. The required individualized assessment implies that the proof and the level of knowledge required may depend on circumstances.

Commercial linkers are expected to “carry out the necessary checks to ensure that the work concerned is not illegally published” and such postings are presumed to “occur with the full knowledge of the protected nature of the work and possible lack of consent”.  [63] The construed knowledge of illegality is therefore based on a duty of care to verify the legality. This level of knowledge can obviously also be strengthened for example through notification.  [64] Inevitably, in some cases it will (later on) turn out that the checks performed have not been sufficient, requiring evaluation of whether the checking still qualifies as a successful rebuttal.

The AG and CJEU presented strong arguments as to why it is difficult to verify the legality of the target site.  [65] These are just the tip of the iceberg. These also affect commercial linkers because the required verification standard cannot be set very high without crippling the operation of the internet.  [66]

The duty of care would suggest requiring some sort of manual assessment when the link is set. Depending on the extent of liability as discussed in Part 4.1, is it required to visit the target before linking, and if so, what precisely would need to be checked? The requisite depth of review is very much an open question. The wording alone implies that there is no need to ascertain absolute certainty and there is no way to do so in any case. It would seem sensible to suggest that the required level of care would be highest with the direct target of the link, might be somewhat lower with the other objects on the target page, and hyperlinks on the target page would require even less verification (if at all). That is, if the link referenced a protected work directly, the level of care would be higher than with more indirect content. Also, the required level of duty of care could also depend on the type of profit-making interests of such linking.

4.3.2. On some practical difficulties of verifying the legality

The first practical problem is that there may be dozens, or even hundreds, of links to third-party websites on a single page, and one may doubt how realistic the verification requirement really is.  [67]

Nonetheless, the evaluation of whether the target is a protected work is complex enough and in practice this would result in erring on the side of caution, with the resulting “chilling effect”.  [68] Such verification could require quite a bit of work and also legal expertise.  [69] Verifying that copyright holders have consented to publication is even more difficult or even impossible: in general, there is not even a way to ensure who is the copyright holder. When it comes to traditional publications, it is not the author, but someone else (a publisher or a third party) who is communicating to the public. The webpage might be unnamed, and the possible works located there might be unnamed as well. How could you even try to determine if these people had consented to publication? Or can you assume that a publisher acts with the author’s consent unless there are strong reasons to believe otherwise?

Article 15 of the Berne Convention and Article 5 of the Enforcement Directive provide presumption of authorship as to who is entitled to enforce rights. These provisions do not apply as such to what linkers are allowed to assume of authorship. In a typical strict liability setting this would be problematic because it would not be enough. The only sensible and practical interpretation of the knowledge and reasonable duty standard is therefore some kind of “obvious illegality” requirement; the “know or ought to have known” standard should not require very extensive verification.  [70]

When considering what would be required of verification, one needs to recall that GS Media concerned deliberate and repeated linking, with headlines resembling “clickbaiting” to obtain financial gain, to obviously illegal unreleased photographs.  [71] The verification standard therefore requires at least that level of care. While the language is generic enough to be read to require also much more extensive care, the arguments on the importance of linking to the internet architecture and freedom of expression clearly suggest that this can be ruled out. However, it remains to be seen where the level of care is set.

4.3.3. Proving sufficient verification and the lack of knowledge

The established duty of care demonstrates a significant problem for all commercial linkers who must be able to rebut the presumption of knowledge. There is no basis to require verification from the non-commercial linkers, although their knowledge of illegality might possibly be established for example if the link itself clearly indicates illegality.  [72] Therefore, when non-commercial linkers are faced with claims of actual knowledge, they may also want to prove the lack of it.

Duty of care could hardly include an obligation to periodically or in some other manner actively check whether the linked page might change. The language in GS Media seems to imply a one-off assessment when the link is set.  [73] Consequently, the events occurring after the link is set should not cause liability. However, actual knowledge established through notification obviously requires reacting in this case.

In practical terms, a strict interpretation of the verification requirements could result in a need to save (as a screen capture or some other way) and keep a diary of such checks to prove what the commercial linker has performed.  [74] Such documentation might prove useful even years later. The evidence of the contents as of the date when the link was set and/or changes afterwards could likely qualify as a rebuttal. The plaintiff may need to demonstrate historical record of illegality; that is, illegality when the link was set. If only proof of current illegality is provided, the rebuttal might succeed by proving that the link was set much earlier.

4.4. Exhaustion or an implied license to link

According to GS Media, as soon and as long as authorized work is freely available, it is allowable to link. It would seem to be impossible to forbid linking or withdraw consent to publish other than by restricting access or obviously removing the content.

An extreme application of the new public criterion has been argued to effectively imply the exhaustion of the right of communication to the public.  [75] The “new public” has also been argued to have been developed as an implicit license argument. Karapapa argues that the implicit license does not cover the act of communication as such, but its intended recipients at a certain point of time, as a consequence allowing the right holders to change their mind later if they so choose. Therefore, it is argued that Svensson is more akin to an affirmation of an implied license.  [76] The argued distinction seems important and convincing. The rejection of the “new public” when it comes to unconsented publications as discussed in Part C.I also supports the doctrinal conceptualization as an implicit license.

Indeed, in Soulier and Doke it was argued that Svensson essentially included a form of implicit consent of the author to link and communicate to the public. [77] Implicit consent typically means that the consent can be somehow changed by an explicit declaration or withdrawal, and the aforementioned judgment concerned precisely that. This could be at odds with the main point of GS Media.

To reconcile these approaches, Svensson and GS Media should not be considered to be based on implicit consent in its purest and unreserved form. Rather, the implicit consent in this setting seems to include mandatory restrictions regarding how it must be withdrawn; that is, by removing or restricting the accessibility of such works, which could be called for example regulated implied consent. [78]

4.5. The conditions of restricting access and circumventing restrictions

What qualifies as an appropriate restriction? According to Svensson and GS Media, the restrictions must be “put in place” or “taken by the site”, they must be used in order to restrict access by the public, and the link must provide an intervention without which others could not benefit from the works. It follows that restrictions must include a technical function that prevents access unless circumvented.  [79] This would preclude for example contractual clauses such as licenses or disclaimers such as “by using this site you agree not to link to it”.  [80]

In consequence, technical restrictions must be implemented in such a manner that existing links no longer allow access. This also avoids problems with the lack of knowledge of changes after the link was set. In practice, adequate restrictions could be performed by modifying the URL and/or using cookies or user authentication. The implication would be that no infringement would be found if the method of restricting access has not at least disabled access using links set earlier.

Is communication to the public when circumventing restrictions based on typical domestic (e.g., strict liability) standard, or are there defenses such as those based on knowledge? Svensson did not discuss any defenses. However, GS Media might have changed the landscape and could be read to provide some support for such argument: ”In contrast, where it is established that such a person knew or ought to have known that the hyperlink he posted provides access to a work illegally placed on the internet, for example owing to the fact that he was notified thereof by the copyright holders, it is necessary to consider that the provision of that link constitutes a ‘communication to the public’ within the meaning of Article 3(1) of Directive 2001/29. – The same applies in the event that that link allows users of the website on which it is posted to circumvent the restrictions taken by the site where the protected work is posted in order to restrict the public’s access to its own subscribers, the posting of such a link then constituting a deliberate intervention without which those users could not benefit from the works broadcast (emphasis added).”  [81]

The first question is whether “the same applies” refers only to the conclusion of the preceding paragraph; i.e., it is communication to the public, or more extensively also to the condition whether the person knew or ought to have known that such posting provides access. The text was probably intended to only cover the conclusion. Nonetheless, the conclusion of the second paragraph, “the posting of such a link then constituting a deliberate intervention”, could be read to implicitly include a requirement that such intervention actually has to be deliberate in some manner.  [82]

It would also be somewhat inconsistent with the rest of GS Media if the user could circumvent restrictions by accident. That is, links circumventing access to legal publications might be evaluated based on the strict liability standard but links to illegal publications based on the actual or construed knowledge standard. It would seem to be preferable to have a similar approach to both cases. While there is no clear basis for a similar “presumption of innocence” for non-commercial linking that circumvents access restrictions, a “knew or ought to have known” standard would not be very far-fetched.

Finally, how noticeable will the restrictions have to be? In the earlier Federal Supreme Court of Germany Session-ID judgment (I ZR 39/08), the restrictions had to be noticeable but not necessarily have efficiency comparable to technical protection measures (TPMs) used in digital rights management (DRM). It would make sense to require at least noticeability in cases to follow. This would also diminish the problems if the aforementioned lack of knowledge defense is not accepted.

5. Conclusions

The CJEU has distinguished linking to consented and non-consented publications in GS Media and Filmspeler both practically and doctrinally from each other.  [83] The former is essentially governed by a new public requirement, which seems to be doctrinally based on a special kind of regulated implied consent, requiring technically restricting access if the author wants to prohibit linking. The latter is essentially governed by actual or construed (based on the due diligence verification requirements) knowledge standard that depends on the profit-pursuing intention and whether the linker has been notified of the illegality.

The most significant open issue – crucial when evaluating linking to non-consented publications – concerns the extent of liability. In addition to the directly linked web page, the linker might also be liable for, for example, at least some clickable links on the linked web page. The interpretation is also very closely related to the duties of care regarding verifying illegality. Taking this to the extreme would lead to impractical and catastrophic results, unless only minimal duties of care were to be adopted. (Un)fortunately the GS Media judgment and its consideration of the factual background was inconsistent and somewhat vague, and therefore both issues are very much open until the situation is clarified by new CJEU judgments.

The interpretation of when posting of hyperlinks is carried out for profit is also a very significant open question. GS Media, Filmspeler, and Ziggo concerned rather severe situations, where the linking and knowledge seemed deliberate and closely tied to pursuing profit.  [84] The CJEU did not specifically say that any economic activity (for example, running a blog with advertisements) would necessarily qualify as such. Further, the degree and type of commercial activity could act as a mediator for the extent of due diligence requirements. That is, the more significant the economic activity is, the more care would be required as to verifying the legality.

In GS Media, human editors seemed to perform linking. It is not obvious that the same principles were meant to apply to automatic linking.  [85] Further, the judgment does not affect intermediaries (such as social media platforms) that operate within the liability exemptions provided by the E-Commerce Directive (2000/31/EC) where links are posted by the users, except that notifications of illegality may also concern links.

Finally, there may be contexts where linking to illegally published material might be socially desirable (for example, journalists describing leaked copyright-protected material). One might suppose there could be exceptions to the GS Media doctrine so that, for example, in some cases the freedom of expression interests could be weightier than intellectual property protection.  [86]

* By Pekka Savola, LL.D. and Lic.Sc.(Tech), Market Court Judge. The views are expressed in a researcher’s capacity. I would like to thank Martin Husovec for the brainstorming and dialogue that led to this article, and Taina Pihlajarinne and the reviewer for helpful comments and feedback.

[1] According to Recital 23 of the directive, “This right should be understood in a broad sense covering all communication to the public not present at the place where the communication originates. This right should cover any such transmission or retransmission of a work to the public by wire or wireless means, including broadcasting. This right should not cover any other acts.” Per Recital 27, the mere provision of physical facilities for enabling or making a communication does not in itself amount to communication.

[2] See e.g. Tsoutsanis: Why Copyright and Linking Can Tango, 9(6) JIPLP 495 (2014), pp. 499–500, 505; ALAI: Report and Opinion on the Making Available and the Communication to the Public in the Internet Environment – Focus on Linking Techniques on the Internet (2013).

[3] Ziggo (Opinion), C-610/15, EU:C:2017:99, para. 4 refers to linking as “secondary communication of works already accessible on the internet” in contrast to “original communication” (in that case, made on a peer-to-peer network).

[4] GS Media, C-160/15, EU:C:2016:644. This topic has also been extensively studied before the latest CJEU judgments, which have in many ways changed the landscape and detailed discussion of earlier accounts is not useful here. For a more detailed account and past criticism, see e.g. Koolen: The use of hyperlinks in an online environment: putting links in chains? 11(8) JIPLP 585 (2016) and Mezei: Enter the matrix: the effect of CJEU case law on linking and streaming technologies 11(10) JIPLP 778 (2016).

[5] For more cases and commentary, see e.g. Clark/Dickenson: Theseus and the labyrinth? An overview of “communication to the public” under EU copyright law: after Reha Training and GS Media – Where are we now and where do we go from here? 39(5) EIPR 265 (2017) and Rosati: GS Media and its implications for the construction of the right of communication to the public within EU copyright architecture, 54(4) CMLR 1221 (2017a), pp. 1233-1237 including the references.

[6] EU Member States are precluded from deviating from the autonomous concept of communication to the public by providing more extensive protection with additional criteria (Svensson, C-466/12, EU:C:2014:76, paras. 33–41). This would also seem to rule out an increased level of protection of similar conduct as such on other grounds.

[7] Also subsequent judgments Filmspeler, C-527/15, EU:C:2017:300 and Ziggo, C-610/15, EU:C:2017:456 have been taken into account to the extent relevant here. (Instead of the CJEU website shorthand “Stichting Brein”, for the sake of clarity, different case names are used).

[8] GS Media (Opinion), C-160/15, EU:C:2016:221.

[9] GS Media, paras. 34–35, which referenced indirectly SGAE, C-306/05, EU:C:2006:764, para. 42 and FAPL, Joined cases C-403/08 and C-429/08, EU:C:2011:631, para. 195. As discussed later, the CJEU does not really address the indispensability requirement later in the judgment.

[10] GS Media, paras. 38–39. These followed Grand Chamber judgment in Reha Training, C-117/15, EU:C:2016:379. In contrast, in Reha Training, para. 49, profit-making was deemed “not irrelevant”. In this context, the further criterion on the number of people in the public would not typically prove to be problematic.

[11] GS Media, para. 41. See Svensson and Bestwater, C-348/13, EU:C:2014:2315.

[12] See GS Media, para 40 and the explanation of Bestwater in GS Media (Opinion), C-160/15, EU:C:2016:221, paras. 36–42. The different means are not germane to this article, however.

[13] GS Media, para. 42, referring to that effect of Svensson and Bestwater. Consent is discussed in Part D.IV. If it was not obvious already with Svensson (see e.g. Headdon: An epilogue to Svensson: the same old new public and the worms that didn’t turn, 9(8) JIPLP 662 (2014), p. 665), inter alia “as long as” seems to confirm that to be relevant, restricting access later must be implemented in such a manner that previously working links cease to function.

[14] GS Media, para. 43.

[15] GS Media, paras. 44–46, in particular para. 46: “Furthermore, it may be difficult, in particular for individuals who wish to post such links, to ascertain whether website to which those links are expected to lead, provides access to works which are protected and, if necessary, whether the copyright holders of those works have consented to their posting on the internet. Such ascertaining is all the more difficult where those rights have been the subject of sub-licenses. Moreover, the content of a website to which a hyperlink enables access may be changed after the creation of that link, including the protected works, without the person who created that link necessarily being aware of it.”

[16] See GS Media, paras. 47–49, 51. Basing the reasoning so heavily on the distinction of profit-making is somewhat surprising, as typically it has little bearing in copyright law (in contrast to, for example, trademarks and commercial use).

[17] GS Media, para. 50, based on Svensson, paras. 27–31. It can be inferred that the restrictions must include some technical function to prevent access, and access may also be restricted later on. See Part D.V.

[18] GS Media, paras. 52. “Another website” has been considered ambiguous (see, for example, Clark/Dickenson (2017), p. 271), and it is. However, at least French, German, Swedish and Finnish language versions use a phrase more equivalent to “some other” or even “any other”, not referring specifically and explicitly to the linked page. The CJEU could have easily used a more specific phrasing if it had wanted to do so. Apparently, a consented publication anywhere on the internet seems to allow linking to non-consented copies. The recent preliminary reference in Renckhoff (C-161/17) should clarify this point.

[19] GS Media, para. 53.

[20] GS Media, para. 53.

[21] Contrast to, for example, the quotation exception in Article 5(3)(d) that allows “quotations for purposes such as criticism or review, provided that they relate to a work or other subject-matter which has already been lawfully made available to the public”. For an implicitly required lawfulness, see for example ACI Adam, C-435/12, EU:C:2014:254, para. 29. Similar kind of support is provided by Filmspeler, paras. 68–69, where the lawful use requirement of temporary copies of Article 5(1)(b) was not satisfied.

[22] See Svensson, paras. 19–23.

[23] See Svensson, para. 24.

[24] In contrast, in GS Media (Opinion), paras. 55–57, the AG argued that the intervention of the hyperlinker must be vital or indispensable, as arguably required in SGAE and FAPL. The CJEU’s argumentation in Svensson and the new public criterion has been subject to a lot of criticism from various fronts. That discussion is however out of scope of this article.

[25] The level of knowledge seems to be the essential qualification, as modulated by the pursuit of profit, not the other way around.

[26] Contrast GS Media, paras. 35, 50 to GS Media (Opinion), paras. 57–60, 69–73.

[27] Likewise, Filmspeler, para. 50, did not dwell on the indispensable role of the seller of the multimedia player, and Ziggo, para. 26 added “or would be able to do so only with difficulty” to the absence of the intervention requirement “those customers would not be able to enjoy the broadcast work”. Midelieva: Rethinking hyperlinking: addressing hyperlinks to unauthorized content in copyright law and policy, 39(7) EIPR 479 (2017), p. 482 argues that in GS Media the CJEU treated the presence of knowledge as negating the lack of indispensability. Rosati: The CJEU Pirate Bay judgment and its impact on the liability of online platforms, forthcoming EIPR (2017b) 3006591 , pp. 3–4 argues that the CJEU has rejected the narrow interpretation of the indispensability of the user’s role.

[28] For example, Rendas: How Playboy photos compromised EU copyright law: The GS Media judgment, 20(11) JIL 11 (2017), p. 14, supported a view that the CJEU’s deliberate nature and profit-making purpose are an alternative to the previous two cumulative elements. More generally, Clark/Dickenson, pp. 272 and 276, have argued that there does not appear to be a coherent, easily applicable overarching theory to cover communication to the public. They have divided the cases to groups concerning retransmission on one hand, and linking and framing on the other.

[29] Filmspeler (Opinion), EU:C:2016:938, paras. 55, 59; Ziggo (Opinion), para. 47.

[30] Filmspeler, paras. 42, 46, 48, 50, 51.

[31] Ziggo, paras. 39, 43, 45.

[32] Ziggo, para. 46.

[33] Ziggo, para. 45, on the new public, references “to that effect” paragraph 50 of Filmspeler, which discusses the knowledge of the provider of the multimedia player that it would provide access to works published illegally on the internet, that is, the deliberate intervention condition of GS Media. It is suggested that this is the finding that Ziggo intended to make, and the preceding paragraph (with its references to Svensson and Bestwater) and the new public discussion were redundant and incorrect.

[34] As criticised by, e.g., Ficsor: GS Media and Soulier – may the hyperlink conundrum be solved and the “new public”, “specific technical means” and “restricted access” theories be neutralized through the application of the implied license doctrine and the innocent infringement defense? (2017), p. 6 and ALAI: Provisional opinion on the right of communication to the public; the Advocate General’s Opinions in Filmspeler Case C-527/15 and Ziggo Case C-610/15 (27.3.2017), pp. 4–5. Indeed, the AG opined in Filmspeler that the subjective component is more appropriate for excluding personal liability than for deciding on objective unlawfulness and the classification of conduct (para. 71). The AG’s opinion was premised on the users merely using hyperlinks without pursuing profit. On the other hand, the CJEU held that the potential purchasers of the specific multimedia player were, as a rule, deliberately accessing unauthorized protected works (para. 69). So, the CJEU did not address this problem in this context.

[35] For example, Article 13 of the Enforcement Directive (2004/48/EC) requires ordering damages as a result of infringement, when the infringer who knowingly, or with reasonable grounds to know, engaged in an infringing activity. However, according to Article 2(1), the domestic law may provide means more favorable for right holders.

[36] In a similar fashion, see Angelopoulos: AG Szpunar in Stichting Brein: An Indirect Harmonization of Indirect Liability. Kluwer Copyright Blog (23.3.2017).

[37] However, there may also be differences, for example, it is not obvious whether and how the level of intention might affect the amount of compensation or damages.

[38] SGAE, para. 42.

[39] See, for example, Angelopoulos: European intermediary liability in copyright: A tort-based analysis (2016); Frosio: From horizontal to vertical: an intermediary liability earthquate in Europe, 12(7) JIPLP 565 (2017), p. 570 fn 61; Midelieva (2017), pp. 484–485.

[40] Ziggo (Opinion), para. 3. Further discussion was provided later, in the event that the CJEU would find that there is no communication to the public based on EU law, at paras. 65–68.

[41] See Angelopoulos (2017).

[42] The existing secondary liability standards and the means of evaluation might provide inspiration in future follow-up cases. At the very least, the courts should try to refrain from adopting interpretations on communication to the public that would be incompatible with established secondary liability doctrines.

[43] GS Media, para. 10: “By clicking on a hyperlink accompanying that text, users were directed to the Filefactory website, on which another hyperlink allowed them to download 11 electronic files each containing one of those photos.” The AG Opinion provides a bit more information (para. 10): “By clicking on a hyperlink, indicated by ‘HERE’, readers were directed to an Australian data-storage website called By clicking on the following hyperlink, they could open a new window which contained the button ‘DOWNLOAD NOW’. By clicking on the button, the readers opened a file in zip format containing 11 files in pdf format, each of which contained one of the photographs.”

[44] GS Media, para. 54.

[45] GS Media, para. 14. The AG did not provide more detail on this (para. 12): “That report, too, contained a hyperlink to the website, where one or more of the photographs in question could be found.”

[46] GS Media, para. 15. The AG provides more context on this (para. 12): “On the GeenStijl forum users then posted new links to other websites where the photographs could be viewed.”

[47] The operative part of the judgment concerns “hyperlinks to protected works, which are freely available on another website” and infringement could be found depending on “illegal nature of the publication of those works on that other website”.

[48] GS Media, para. 51, also in a similar fashion at para. 53.

[49] Clark/Dickenson (2017), pp. 269–270.

[50] Rendas (2017), p. 14 fn 38; Rosati (2017a), p. 1238.

[51] See e.g. Rosati (2017a), p. 1238. See also Brüss: Hamburg court applies GS Media for the first time in Germany, 12(3) JIPLP 164 (2017); Malovic/Haddad: Swedish court finds that an embedded link to unlicensed content infringes copyright, 12(2) JIPLP 89 (2017). Cf. Leistner: Closing the book on the hyperlinks: brief outline of the CJEU’s case law and the proposal for European legislative reform, 39(6) EIPR 327 (2017), p. 331 criticizes domestic courts interpreting GS Media too literally and without taking into account what is reasonable and proportionate.

[52] GS Media, para. 47: “posting of a hyperlink to a work freely available on another website is carried out by a person who, in so doing, does not pursue a profit” and para. 51: “posting of hyperlinks is carried out for profit”.

[53] Papasavvas, C-291/13, EU:C:2014:2209, para. 30: ”Article 2(a) of Directive 2000/31 must be interpreted as meaning that the concept of ‘information society services’, within the meaning of that provision, covers the provision of online information services for which the service provider is remunerated, not by the recipient, but by income generated by advertisements posted on a website.”

[54] See the description earlier in Part D.I.1.

[55] GS Media, paras. 54, 7: ”GS Media operates the website GeenStijl, which includes, according to information provided by that website, ‘news, scandalous revelations and investigative journalism with lighthearted items and wacky nonsense’ and which is viewed daily by more than 230 000 visitors, making it one of the 10 most visited websites in the area of news in the Netherlands.”

[56] In a somewhat similar fashion, Clark/Dickenson (2017), p. 269, argue that the CJEU attempted to distinguish average internet users and deliberate profit-seeking infringers.

[57] Likewise considering also blog operators and advertising, see Rendas (2017), pp. 14. Cf. Leistner (2017), p. 330 deems “any posting on websites refinanced by advertisements would undoubtedly have to be regarded as postings carried out for profit”.

[58] On predominantly minimal profits from advertisements, see Midelieva (2017), p. 484 fn 91.

[59] See Clark/Dickenson (2017), p. 270 on knowledge and profiting forming a kind of a circle.

[60] C-324/09, EU:C:2011:474, paras. 120, 122.

[61] Likewise, Leistner (2017), pp. 330–331 calls for adjusting, specifying and reasonably limiting duty of care, taking into account all the specifics of the case in light of the proportionality principle.

[62] For example, Leistner (2017), p. 331 describes the facts of GS Media as very specific, including multiple warnings on infringing links and persisting in infringing activity nonetheless.

[63] It is not obvious why the judgment mentioned “possible” lack of consent, but not when it comes to the protected nature of the work. One explanation might be that the linker cannot know for sure whether publication has been consented to, given (for example) that the same work could have been published freely accessibly with consent somewhere else.

[64] What kind and how specific notice is required to establish knowledge, for example when compared to current Notice-and-Takedown regimes? Because there are no required formalities, more informal notice could be adequate. It seems plausible to require identifying the specific URLs or otherwise providing specific information where the link that should be removed is located. Specific claims and the basis for illegality should obviously also be provided. The implications of insufficiently substantiated notification are also open to debate.

[65] See GS Media, para. 46 and GS Media (Opinion), para. 78.

[66] In a similar fashion, see Leistner (2017), p. 331.

[67] See, e.g. Rosati (2017a), p. 1232 and the example provided therein.

[68] In a similar fashion on the chilling effect, see Rendas (2017), p. 15.

[69] On earlier discussion of verification requirements, see e.g. Schellekens: Reframing hyperlinks in copyright, 38(7) EIPR 401 (2016), p. 404. Also on difficulties, see Leistner (2017), p. 331.

[70] In a similar fashion, Leistner (2017), p. 331 argues for “compliance with minimal, most basic, duties” which could in certain automatic linking contexts be even close to non-existent.

[71] Likewise, the intention and knowledge were apparent in Filmspeler and Ziggo.

[72] For example, linking to a “” subpage including a protected work could be argued to be in scope of actual knowledge also for non-commercial linkers.

[73] In GS Media, para. 51, the necessary checking requirement and the status of legality are connected to the time of posting of the hyperlink. While discussing the challenges of verification, the reasoning also includes text to the effect of “may have changed in the meanwhile”, which if read out of the context might suggest liability could result in such a case at least in some circumstances. Rosati (2017a), p. 1232, also argues the impossibility of making sure that “links provided are and remain (the content linked to and hosted on a third-party website can in fact change over time) to licensed content”, possibly implying an ongoing responsibility. However, the context of discussing difficulties of too broad interpretation should not be taken to imply affecting liability standard, but rather to the contrary.

[74] A cautious commercial linker might in any case need to adopt this methodology until the requirement has been clarified.

[75] For example, Hugenholz/Van Velze: Communication to a new public? Three reasons why EU copyright law can do without a “new public”. 47(7) IIC 797 (2016), p. 811. Even more forcefully, see e.g. Rosén: How Much Communication to the Public is ‘Communication to the public’?”, in Stamatoudi (ed.): New Developments in EU and International Copyright Law. Kluwer Law International (2016), pp. 331–350.

[76] Karapapa: The requirement for a “new public” in EU copyright law, 42(1) ELR 63 (2017), pp. 64, 74–75. For an argument for applying implicit license before Svensson, see Pihlajarinne: Setting the limits for the implied license in copyright and linking discourse – the European perspective, 43(6) IIC 700 (2012).

[77] Soulier and Doke, C-301/15, EU:C:2016:878, para. 36, where “in a situation in which an author had given prior, explicit and unreserved authorisation to the publication of his articles on the website of a newspaper publisher, without making use of technological measures restricting access to those works from other websites, that author could be regarded, in essence, as having authorised the communication of those works to the general internet public.” Cf. for example Ficsor (2017), p. 11, seems to criticize whether the consent is sufficient merely due to the lack of technological protection, also accepting forbidding linking in contractual terms or through notice on the website.

[78] Alternatively, the theory could be conceptualized as a mandatory license of a sort: keeping works freely available on the internet inherently includes certain restrictions.

[79] In a similar fashion, Headdon (2014), p. 668 and Koolen (2016), p. 594.

[80] Cf. Ficsor 2017, p. 11, who includes contractual terms or notices and Mezei (2016), p. 782, who included also paywall registration pages or robots.txt files as sufficient restrictions.

[81] GS Media, paras. 49–50.

[82] Likewise, “full knowledge of consequences” has been taken as a sign that some form of knowledge is required for liability in Schellekens (2016), p. 404.

[83] As discussed in Part C.I, Ziggo could be considered a misstep or an application of communication to the public in general, rather than linking in particular.

[84] Quite another issue is that the defendants may not have anticipated that linking by them or the third parties could lead to their liability for copyright infringement in the first place.

[85] This is particularly significant when it comes to the verification requirements, which seem to be premised on human and visual inspection of the linked website. For example, Leistner (2017), p. 331 fn 39 considers limiting search engines and aggregators’ reasonable checks to metatag automation.

[86] For discussion, see, for example, Geiger/Izyumenko: Copyright on the Human Rights’ Trial: Redefining the Boundaries of Exclusivity Through Freedom of Expression, 45(3) IIC 316 (2014) and the referred European Court of Human Rights judgment Ashby Donald v. France of 10.1.2013 (case 36769/08).



Any party may pass on this Work by electronic means and make it available for download under the terms and conditions of the Digital Peer Publishing License. The text of the license may be accessed and retrieved at

JIPITEC – Journal of Intellectual Property, Information Technology and E-Commerce Law
Article search
Extended article search
Subscribe to our newsletter
Follow Us