The New Copyright Directive: Article 17 and copyright limitations – picking two cherries and leaving the rest to spoil? Part I

Kluwer Copyright Blog
October 29, 2019

Please refer to this post as: , ‘The New Copyright Directive: Article 17 and copyright limitations – picking two cherries and leaving the rest to spoil? Part I’, Kluwer Copyright Blog, October 29 2019, http://copyrightblog.kluweriplaw.com/2019/10/29/the-new-copyright-directive-article-17-and-copyright-limitations-picking-two-cherries-and-leaving-the-rest-to-spoil-part-i/


Article 17 of the adopted DSM Directive requires that so-called online content-sharing service providers (OCSSPs) either obtain use licenses from rightholders or, failing that, enforce copyright ex ante by preventing uploads. At the same time, according to Article 17(7) any agreements between rightholders and OCSSPs cannot affect the availability of content created under the limitations, and Member States must ensure that users are able to rely on certain existing limitations. Furthermore, Article 17(9) introduces certain procedural safeguards for being able to rely on limitations. Thus Article 17(7) and (9) DSM Directive establish an interface between OCSSPs’ obligations and the European copyright limitations framework (Article 5 InfoSoc Directive). These two posts address this interface.

Article 17(7) contains two parts, each of which focuses on different parties. The first paragraph explains the treatment of limitations in potential arrangements between rightholders and OCSSPs. The second paragraph introduces certain obligations for the Member States. In this first post I will focus on the former aspect.

A new framework for European limitations

The harmonisation of limitations has been a thorny issue in European copyright law. Article 5(2)-(3) InfoSoc Directive provides a basket with cherries from which Member States are allowed to pick and choose from an exhaustive list of 21 different provisions. Only one cherry must be picked and implemented, a limitation to the reproduction right allowing temporary copying (Article 5(1)). According to the adopted wording, the basket is covered by a European version of a three-step test (Article 5(5)) which must be satisfied before any of the provided limitations are applied. Furthermore, the limitations have been held to constitute exceptions to a general rule, for which reason they must be interpreted strictly (Infopaq, para 56). Finally, although recital 31 InfoSoc Directive makes it clear that one of the Directive’s objectives is to safeguard a fair balance of interests between rightholders and users, it does not really explain what those interests are. On the rightholder side, recital 9 states that a high level of copyright protection is expected.  For any other interests, the Directive merely emphasises the need to promote learning and culture (recital 14), although not by sacrificing the strict protection of rights (recital 22). In many ways therefore, the limitations framework, as it was adopted in 2001, has lain within the domain of the Member States.

The increasing number of references to the CJEU on copyright has allowed the Court to play a pivotal role in dynamically harmonising a great portion of the framework. Although Article 5 InfoSoc Directive remains a cherry basket, the three-step test is not always applied by the CJEU (e.g. Deckmyn), other times the test is presumed to apply (e.g. FAPL, paras 181-182), and the Court has moved from a strict interpretation of limitations to a teleological interpretation (e.g. Painer, para 133, Ulmer, para 32, 43), recently even recognising that the limitations confer rights on their beneficiaries (Spiegel Online, para 54,  Funke Medien, para 70). While it remains to be seen what the latter means in practice, a clear trend of giving well-deserved and desirable attention to the limitations is materialising. This is reinforced at legislative level with the binding character of the Charter of Fundamental Rights, whose adoption formally injected constitutional values into the framework and lifted copyright balancing to the level of fundamental rights discourse. As such, most of the limitations in the adopted exhaustive catalogue of Article 5 InfoSoc Directive are now clear embodiments of protected fundamental rights also in EU, as opposed to only national, copyright law. Those rights and values have come to play their role not only in the context of the limitations, but also in the context of enforcement of copyright (e.g. C-70/10 Scarlet Extended).

In sum, Europe is approaching, but is yet to arrive at, a copyright limitations system proper. As a result, the new obligations for OCSSPs under Article 17 DSM Directive bring about difficult tensions between the recognised interests of all the parties affected by Article 17.

Ensuring the availability of non-infringing content on platforms

The first paragraph of Article 17(7) seeks to ensure that the co-operation between OCSSPs and rightholders (which under Article 17(2) ought to take the form of licence agreements in the first place) does not result in the prevention of the availability of content uploaded by users which does not infringe copyright or neighbouring rights. This expresses an obvious public policy choice, since there is no copyright interest over content that does not infringe rights. Beyond that, curiously, the effect of the provision seems to be to prevent the contractual disposition of the availability of non-infringing content.

Works and subject matter “covered by an exception or limitation”

According to Article 17(7), the term ‘non-infringing’ includes works and other subject-matter that are “covered by an exception or limitation”. The reference to works and other subject matter, as opposed to the use of works and other subject matter, perhaps implies that the uploaded content must have been created on the basis of a limitation, i.e. it must involve some kind of transformative use. This would exclude most limitations, since very few actually require transformation of the original for the limitation to apply. What is relevant in most cases is the purpose for, or context in which content is used. Transformation of protected content is only required in 5(3)(k) concerning parodies, caricatures and pastiches, and (possibly) Article 5(3)(d) concerning quotations. This appears to be consistent with recital 70 which states that “users should be allowed to upload and make available content generated by users for the specific purposes of quotation, criticism, review, caricature, parody or pastiche”.

Subjecting this part of the provision to such a narrow interpretation means that the cooperation between rightholders and OCSSPs is only prevented from rendering unavailable content which has been created on the basis of two limitations. Thus, it does not guarantee the availability of content which otherwise may be made available on the basis of other limitations, such as Article 5(3)(i), which permits the incidental inclusion of a work or other subject matter in other material. This awkward balancing, which effectively valuates the limitations, is, according to recital 70, grounded in users’ freedom of expression interests. Yet why incidental inclusion of protected content in other material is not on this list is not easily understood. Does it fall outside freedom of expression? Perhaps it is an accidental oversight, reflecting only the complexity of the negotiations and not the legislator’s intention, like the excerpt from recital 70 quoted above, which seems to limit itself to allowing users to reuse user-generated content (similar sentences appear in the Swedish and Polish language versions, although these state instead that users should have a right to upload and make available such content).

Non-infringing, but not covered by an exception or limitation

The alternative to being covered by a limitation yet remaining non-infringing, which is not mentioned in the text of the provision, includes the following situations:

  1. when protected content is used on the basis of a licence,
  2. when previously protected content has fallen into the public domain and may be used freely,
  3. when the work or uploaded parts thereof is not original,
  4. when the user-generated content (which often must become a work) transforms the earlier work to such a degree that it is freed from any interest of the original rightholder (e.g. 4§ 2 st. Swedish Copyright Act, §24 Abs. 1 German Copyright Act),
  5. when the user-generated content incorporates and transforms subject-matter protected by neighbouring rights to such a degree that it becomes unrecognisable to the human senses (Pelham).

Except the first two points above, none are capable of being established with certainty otherwise than following a decision of a competent court or tribunal. Instead, all three are likely to give rise to diverging views. The same is true for determining the applicability of a limitation to the use of content which is protected.

Assessing the (non-)infringing status of content

As a matter of bargaining power and risk assessment, diverging views over the status of content may ex ante tilt the balance in favour of rightholders, unless there is to be a change of the burden of proof. Article 17(9) does require that rightholders duly justify their requests for disabling access or removing the content and such requests shall be subject to human review, which permits context-specific assessments. The CJEU’s recent recognition of the limitations as conferring rights on its beneficiaries perhaps makes such an unorthodox understanding more plausible. On the other hand, the fact that Article 17(1) states in absolute terms that OCSSPs communicate content to the public, combined with the inapplicability of the safe harbour provision in Article 14 E-Commerce Directive (Article 17(3)), removes incentives to fully scrutinize the existence and scope of protection, including the scope of any applicable limitations, since it exposes OCSSPs, the first instance deputised decision-makers, to liability for copyright infringement. Thus, although the co-operation between rightholders and OCSSPs in an ideal world should not prevent the availability of non-infringing content, deciding whether content is non-infringing is a game that may come to involve high stakes that solely depend on the degree to which OCSSPs are prepared to defend themselves in court.