EU Proposal for a Regulation Preventing the Dissemination of Terrorist Content Online: An Overview

The Commission announced yesterday a proposal for the Regulation Preventing the Dissemination of Terrorist Content Online. Somebody not following the developments closely might get an impression that this is a stand-alone initiative. In reality, the Proposal follows both a broader drive to regulate platforms, announced in the 2015 DSM Strategy, followed up in the 2016 Communication on platforms, and further elaborated in ‘soft law’ 2017 Communication and 2018 Recommendation on illegal content online.

The desire to regulate “platforms” is, in itself, problematic. Platforms are neither natural subjects for IT regulators (unlike telecoms networks & services or information society services), nor sufficiently clearly defined to lend themselves to straightforward regulation. They differ in size, scope, type and impact and use a vast plethora of business models. The EU’s continuous drive to regulate them without exploring deeper implications of such approach is worrying.

The present Proposal aims to “prevent the misuse of hosting services for the dissemination of terrorist content online”. It imposes duties of care on hosting services to prevent dissemination of terrorist content and measures on Member States to identify the content and remove it. The Proposal has the same wide scope as GDPR, applying to all hosting providers targeting services in the Union, irrespective of their place of establishment.

The definition of terrorist offences is taken from the 2017 Directive on terrorism. Terrorist content, which is the Proposal’s main target, is defined as inciting, advocating, encouraging, promoting or instructing on terrorist offences so defined in the 2017 Directive. Hosting service providers are obliged to take action against dissemination and include provisions to that effect in their terms and conditions.

The interesting (and controversial) part of the Proposal appears in the form of “removal orders” of Article 4. These are non-voluntary demands issued by competent authorities and directed at hosting service providers. National authorities are allowed to require providers to remove content and disable access to it and the latter would be required to do so “within one hour from receipt of the removal order”. This obligation seems to be applicable irrespective of the size of the site or the hours under which it is manned. A statement of reasons is available but only upon request from the provider and cannot, in any case, delay the removal order. If the hosting provider disagrees with the order due to “manifest errors” or because it needs “clarification”, the removal is postponed until such clarification is provided. In addition to the removal orders, the authorities may send voluntary requests called “referrals” (Article 5) which providers are free to assess against their own terms and conditions but need not remove.

Another controversial feature are “proactive measures” that Article 6 demands hosting providers make. The providers need to take “effective and proportionate” measures, “where appropriate”, while assessing risks and taking fundamental rights into consideration. Once a removal order of Article 4 has been issued, though, special proactive measures that relate to the hosting provider which has been the subject of such order kick in, demanding that they submit annual reports about prevention of re-upload and detection, removal and disabling of content. Where deemed insufficient, further proactive measures may be required and imposed forcefully.

Article 8 demands transparency from hosting providers, including the use of terms and conditions while Article 9 requests human oversight of automated removal measures. The rest of the Proposal contains detailed measures on complaints, cooperation, implementation and enforcement.

Particularly interesting is the status of hosting providers located outside the EU. They are, under Article 16, required to designate a legal representative for the purposes of compliance. Under Article 15, where a provider does not have a place of establishment in the EU, it is the place of residence or establishment of the legal representative that is the place relevant for enforcement purposes.

The penalties set out in Article 18 are for Member States to determine but are obliged to punish systematic failures by 4% of the hosting provider’s global turnover.
There are, in my view, three main problems with the proposal:

  • first, the Proposal derogates specifically (Recital 19) from the obligation not to monitor, set out in Article 15 of the E-Commerce Directive. This is a dangerous and poorly justified precedent. Although the recital does speak of “balancing” in such cases, it seems as if a mere fact that a label “terrorist” had been stuck on particular content, at a particular provider, would trigger a derogation the effect and duration of which is unknown and not justified or considered by Article 15.
  • second, complying with obligations in the Proposal, and in particular Article 6, will require the use of filters, since relying on human moderators would be disproportionately expensive. Filtering technology works sporadically and tends to ‘catch’ legitimate content, often opening more problems than it solves.
  • third, the cost-benefit side of the question is obscure in the proposal. Terrorist content tends to appear and disappear quickly and is rarely simply placed on platforms, for them to conveniently remove. It tends to shift from one account to another and from one platform to another. Burdening the hosts with 24/7 removal and monitoring obligations may look justified but is unlikely to give desired results and may, in worst cases, lead to governmental abuse. It is further unclear whether non-EU providers would simply withdraw from the EU (as some did as a result of GDPR’s extraterritorial reach).

In summary, the problem is more complex than the regulator would like to believe and may require creative thinking, the use of co-regulation and technical solutions of the next generation rather than heavy-handed removal and penalty system.

A comment on the JURI Committee Report on Article 13 of the Copyright Proposal and everything that is wrong with it

Silke von Lewinski published today a comment on the JURI Committee Report of the Copyright Proposal. I am reposting here what I posted as a comment to her original post and should be read as such.  I believe that it is important to keep the momentum going as we approach further debate in the Parliament.

First, that platforms are legitimate targets of regulation is by no means a foregone conclusion. That the proposal targets “platforms” rather than “information society services”, and makes a number of assumptions on the way, is a problem of enormous proportions, not just in terms of how well the proposal communicates with the E-Commerce Directive, but also in terms of is relation to other EU IT laws. In short, one cannot just wave a magic wand, assume that all UGC cites are “platforms”, and then happily continue discussing whether such platforms engage in communication to the public, as if this move had no consequences, all of this in almost complete absence of an official EU policy on platforms. This is an issue I discuss in a forthcoming article, (2018) 6 CLSR.

Second, the Proposal simply states that “active” online platforms perform a communication to the public. What an “active” site is has been debated ad nauseam for over two decades now, but is still not a point which has been adequately resolved in CJEU case law, which is obscure and can often only be read in its particular context. One cannot equate TV sets in hotel rooms with streaming services or cloud depositories. Neither can one use Pirate Bay – an openly and proudly illegal distributor – as a comparison point for a streaming/distribution platform, no matter how active the latter might be. This problem is about communication to the public of a bona fide platform, and Pirate Bay case is of limited to no use here.

Third, the main problem in this issue is not whether a “platform” (for what this term is worth) can technically be deemed to fit within various bits and pieces of law on communication to the public but whether it makes sense to dump all online distributors into the same category and burden them with regulation and filtering obligations in almost complete absence of harmony with the spirit and letter of ECD Articles 12-15. The ISP liability regime has been designed with a very specific idea of avoiding indiscriminate filtering and cannot be changed by simply labelling all online distributors as active – a presumption of huge consequence.

Fourth, while there is no doubt that Article 14 was never meant to protect knowledgeable providers or primary infringers, the Voss proposal sharpens the already muddled Preamble 38 by making what is essentially an irrebuttable presumption of knowledge for all UGC ISPs. Whereas the original Proposal talks of communication only where there are activities “beyond the mere provision of physical facilities”, the Voss text simply states that all online content sharing providers are communicators. Why would they be so and, even where they are, why would they exercise prior rather than subsequent restraint? I will not even begin on the subject of how broad the definition of 37a really is. This subjects a vast and ill-defined category to a poorly designed and heavily-lobbied-for set of expensive filtering. Surely, this cannot have been the intention of the original drafters of InfoSoc Directive?

Fifth, the main problem of the both the original Proposal and the Voss text is that indiscriminate and expensive filtering of the kind only Google can pay for is demanded. All this in complete contradiction to SABAM case which is very clear on this (but is not discussed in the article above).

Finally, the Proposal is incredibly badly harmonised with the spirit and letter of the E-Commerce Directive. I am afraid that no amount of careful interpretation of the communication to the public will resolve this. The Commission has been told, a vast number of times, by very many people that this is a problem of gigantic proportions. It has so far chosen to ignore it, which fortunately led to its defeat in the early summer. Let us hope that the Proposal will be defeated once and for all in September so that we can begin a serious and much-needed discussion on the reform of EU copyright law.

The Commission’s Proposal on Fairness and Transparency for Business Users of Platforms – Online Platforms to Be More Transparent, Disclose Ranking Criteria

On April 26, the Commission published a proposal for a Regulation on promoting fairness and transparency on platforms. This proposal follows the 2015 EU Digital Single Market Strategy promise to look into platforms and a Communication on Online Platforms the Commission published in 2016. While these documents did not promise an overarching law on platforms (covering both the B2C and B2B situations), they signalled the Commission’s willingness to look at a set of targeted issues and introduce legislation if and where necessary. The present proposal aims to improve transparency in B2B relations involving platforms on one side and businesses which otherwise provide services to other businesses and consumers through such platforms on the other. The main idea is that platforms act as “gatekeepers” of the online world, effectively bringing dependency for many businesses. The present proposal brings extra requirements for clarity and transparency in situations where platforms could abuse their dominance. It seems that the choice of transparency (rather than blacklisting certain practices) as the Commission’s main tool is meant to fulfil proportionality and subsidiarity criteria, leaving further action to Member States and/or competition and marketing laws. The Proposal is not meant to be full harmonisation.

The regulation would apply to online intermediaries and search engines (hereafter “platforms”) providing services to business users with a place of establishment in the EU and offering goods and services to EU consumers. It is irrelevant for the purposes of the Proposal if platforms themselves have a place of business or residence in the EU, as long as business users they offer services to have such a place and target EU consumers.

An intermediation service, which is in the Proposal placed under transparency and fairness obligations, is defined as a) an information society service (ISS) within the meaning of the E-Commerce Directive, b) that facilitates direct transactions between business users and consumers and c) which are provided to business users on the basis of an underlying relationship between online platforms and others (businesses and consumers). A provider can either be a natural or a legal person. Online search engines are defined simply as digital services allowing users to perform search. The definition of ISSs seems very broad as these are, in reality, all electronic services provided at distance and with remuneration (the latter not necessarily in the form of a payment). Additionally, the majority of ISSs do facilitate B2C transactions.

The obligations to be introduced can be summarised as follows:

  • Article 3 requires that platforms increase transparency of their unilaterally drafted terms and conditions, and in particular any modifications made to them.
  • Article 4 requires that any suspension and termination of platform services be accompanied by a statement of reasons.
  • Article 5 brings potentially the most significant changes. Online platforms would be required to set out in their terms and conditions the main parameters determining ranking and the reasons for their relative importance. Additionally, where businesses have an opportunity to pay to influence ranking, such possibility would also have to be disclosed. Online platforms are not required to disclose any trade secrets as defined in EU Trade Secrets Directive.
  • Article 6 obliges intermediaries to disclose in their terms and conditions any preferences given to goods and services they themselves (or businesses they control) offer. This covers situations where platforms offer and promote their own prducts, as was the case with Google comparative shopping.
  • Article 7 requires the disclosure (description) in terms and conditions of any use of data that business users and consumers disclose to platforms.
  • Article 8 requires that platforms that prohibit provision of same good and services through other means than through platforms to disclose such restrictions. This covers situations where platforms demand exclusivity from their business customers.
  • Article 9 introduces an internal system for complaint-handling.
  • Articles 10-11 introduce the possibility of using mediation.
  • Article 12 allows organisations with a legitimate interest representing business users and public bodies in Member States to take action to stop or prohibit non-compliance. Organisations in question need to be non-profit.

While transparency requirements in some of the articles seem superficially reasonable, I see two major problems with the proposal.

The first is its extent. While there is some agreement as to what constitutes a search engine,1 the same cannot be said of intermediaries. Even if one assumes that most intermediaries that fall under the E-commerce Directive would also fall under the Proposal, there is still scope for questioning the reasonableness of such proposal.

The second problem is the potentially very wide scope of Article 5. Ranking criteria are not determined on a whim but are a result of algorithms which are a business asset and a trade secret. While the proposed Article does not require disclosure of anything that the Trade Secrets Directive itself does not consider a trade secret,2 it seems to be impossible to release meaningful information on search ranking criteria without also releasing trade secrets.

It remains to be seen whether the proposal will move smoothly through the legislative procedure. This seems very unlikely, as heavy lobbying is expected from a variety of platforms. This brings into question the meaningfulness of the Commission’s exercise. Under the circumstances, a more vigorous application of competition law would have probably achieved the same result.

  1. Although no full agreement. Is Facebook (also) a search engine? Instagram?
  2. Secret and not generally known, with a commercial value and subject to controller’s reasonable steps to keep it secret.

Bad Reporting of EU Tech Policy and Regulation in the Media: How to Recognise it and How to Get the Real Picture

While there can be no doubt that EU tech policy cannot be considered the sexiest topic on the planet, it is nevertheless covered very frequently in the media. In the era of privacy leaks, government surveillance, AI and streaming, to name but a few, it is also very relevant. It is not surprising, therefore, that reputable newspapers often cover real and imagined EU activities with vigour. A careful reader would, however, find that articles written by experienced journalists often carry sensationalist titles. Even more surprisingly, such a reader would have difficulties confirming the article’s claims on EU’s own portals. The reporting is often incoherent, confuses different policy areas and rarely quotes original sources or points the reader in the wrong direction. This was particularly obvious in the recent wave of coverage of global platforms such as Google and Facebook. Here are some of the examples:

On March 14, 2018, Reuters carried a story with a bombastic title: “Google, Apple face EU law on business practices”. The article claims that the EU is “drafting a new regulation”, that this regulation will be “specifically targeting online platforms” and that it will ask companies to be more transparent on how they rank search results. The same story was subsequently carried, almost verbatim, by a number of newspapers and online portals, including the Financial Times, the EU Bulletin and Business Review. Anybody being even vaguely familiar with the EU’s Digital Single Market project would be surprised at such statements. This is more so since the Commission’s own 2016 Communication on Platforms states that no new directive or regulation on platforms is forthcoming nor does it list search engine ranking algorithms as being under scrutiny. While the Commission is not organised in the way it presents its activities, it rarely acts without proper announcement.

What is the origin of the story, then? The mystery is not very deep. A careful reader would notice that a new EU package on consumer law appeared on April 11. Part of that package is a proposed Directive updating a bunch of EU consumer laws (informed, among other things by a behavioural study on transparency in online platforms), including the 2005 Unfair Commercial Practices Directive. Suggested as one of the changes is the new Article 6a, applying to online marketplaces. It says that “the main parameters determining ranking of offers presented to the consumer as result of his search query on the online marketplace” must be disclosed to consumers or the practice would be considered an unfair one. More interesting, however, is the Commission’s work on fairness in platforms-to-business relations. As part of the 2015 Strategy’s promise to look at platforms, the Commission promised to look at transparency of trading practices of online platforms. The inception impact assessment, published in late 2017, looks at three legislative options: soft law, targeted EU instrument coupled with industry-led action or detailed EU principles. Whoever wrote the original Reuters article would have seen the work on the consumer package and/or seen the transparency document on fairness in platform-to-business relations and may have also seen a draft document as an outcome of one of the three options. They then constructed the story which, while containing elements of truth, is still considerably off the mark. It is also worth noting that even modestly controversial EU proposals get significantly modified and are occasionally withdrawn where the Council and the Parliament cannot reach an agreement. What will happen, at best, is that a proposal will be tabled and discussed at length.

Another article appeared in Financial Times on April 17. It claimed that EU is “to give judges power to seize terror suspect emails and texts”. It begins with the familiar “Brussels is planning” words and goes on to say that judges will get “the power to seize emails and text messages of terror suspects” and that the Commission “will propose giving national judges the extraterritorial power to order companies to hand over ‘e-evidence’ held in servers in another EU country or outside of the bloc.” While this sounds both dramatic and controversial, readers would be hard pressed to understand which regulatory platform this proposal falls under, which DGs might be involved, where to find the policy document, how to follow the developments or how this proposal fits in the broader Digital Single Market Strategy. A much more coherent (albeit shorter) account of the affair is to be found on Politico’s website. There, it is clarified that the proposal is part of the Commission’s drive to improve taking of e-evidence and outlines its main features as well as differences from the current regime. Crucially, the article identifies that this is part of the effort to achieve a “common judicial area” and that it only applies to serous crime. A casual look at the Commission’s website on taking cross-border evidence would reveal that this is indeed part of an ongoing effort in cross-border judicial cooperation, that “European Investigation Order” is already in force and that work continues on the “European Production Order”. While the FT article is not entirely wrong in its reporting, the value it adds to the debate is minimal.

Similar stories are to be found all over the Internet. It is usually easy to recognise them as they are rarely specific and are usually framed in the clichéd “Brussels plans” or “EU threatens” style. How is the reader to find out what is really behind the story?

The first step is to identify the area the tech policy belongs to. Very roughly, three regulatory circles exist in the EU’s Digital Single Market effort. Information society services are regulated under the e-commerce framework (here also including copyright, privacy, speech regulation, etc.) Telecoms regulatory framework covers telecoms networks and services. Finally, media under editorial control are subject to Audio-Video Media Services Directive (AVSMD). The circles, although distinct, overlap. Privacy, for example, is subject both to e-commerce rules (GDPR) and telecoms rules (ePrivacy regulation) which means that a full picture can only be obtained by looking at both.

The second step is to identify the general EU policy in the area in question. The 2015 Digital Single Market Strategy should always be consulted as should any other statements that further clarify the political EU position of a particular issue. For platforms, for example, there exists the Communication (mentioned above), but also the 2017 Communication and the 2018 Recommendation on Illegal Content online. The formation of a potential proposal is always followed by various studies the Commission publishes (usually outsourced to think tanks) as well as stakeholder consultations, speeches and announcements on web sites. The platform law above, for instance, is preceded by the already-mentioned inception study and the behavioural study on transparency.

The third step is to check whether any policy ideas identified in steps 1 and 2 have been transformed into legislative proposals. Privacy, consumer protection, cybersecurity, copyright, telecoms, AVMS services and e-commerce all have their own websites (sometimes multiple ones at different institutions). These not only summarise present laws but explain policies and plans for future regulation. Further to that, stakeholders are often engaged in commenting on relevant proposals, both officially (as part of the Commission’s fact-finding exercise) and in their own reports, blogs, etc. Such documents are invaluable for gauging public and professional opinion. The final step would be to confirm that proposals are about to become laws. Existing proposals can be traced through the Legislative Observatory.

While the above may seem banal, it is rarely followed, which results in confusing and messy reporting. Although it may sometimes seem that proposals come “out of the blue”, this is in reality almost never the case as significant preparatory work is needed and the Commission launches initiatives based on plans announced in strategy documents. In other words, the Commission maintains a certain degree of transparency for political and budgetary reasons and has no political mandate to randomly tackle issues not otherwise announced as part of a wider political agenda. In such circumstances, it becomes very important to place the news in their proper context.

The EU Reform of Consumer Law: A Quick Overview of the “New Deal”

After promising a comprehensive revision of consumer law in September 2017, and in light of the fitness check published in May 2017, on 11 April 2018 the Commission published the New Deal for Consumers. In justifying the need for the reform, the Commission states that, while the current substantive rules are “fit for purpose, their effectiveness is hindered by lack of awareness and by insufficient enforcement and consumer redress opportunities.“ The new package is mainly designed to stop and deter infringements (more effective prevention) and ensure redress when needed (better enforcement).

The present EU consumer law framework consists of a patchwork of rules dating to different periods. It has largely not been touched since the 2011 Consumer Rights Directive had been adopted.1 While this framework functions relatively well,2 it has not been subject to serious updates for eight years. In the meantime, new challenges have arisen, including an unprecedented rise in online sales, digitization of the society and more aggressive profiling and tracking. The New Deal brings the framework in line with the new digital reality.

The reform consists in a Communication and two proposals for Directives:

  • Communication on A New Deal for Consumers. This document simply summarises the status quo, states the objectives and explains what each of the two directives do.
  • Proposal for a Directive on representative actions for the protection of the collective interests of consumers. This directive introduces a collective redress right in cases where groups of consumers have suffered harm. Consumer organisations would, under this proposal, be able to demand redress.
  • Proposal for a Directive on better enforcement and modernisation of EU consumer protection rules. This proposal aims to modernise consumer law, in particular enforcement. It does so by modifying the 1993 Unfair Terms, the 1998 Consumer Protection, the 2005 Unfair Practices and the 2011 Consumers Rights directives.

The changes that the proposed directives bring can be summarised as follows:

  1. In terms of the new collective redress: Qualified entities will be able to commence actions representing consumers’ collective interests. The proposed Directive is not a full harmonisation, leaving Member States the opportunity to introduce other collective remedies. The Directive would apply to infringements of EU law involving consumers’ collective interests and listed in the Annex. Collective interests are defined simply as those affecting multiple consumers (e.g mass recalls, delays, faults, etc.) The Directive does not change Member States’ laws regarding other available actions, nor does it affect private international law issues (other than to make class actions possible). Qualified entities are to be designated by Member States.
  2. 2005 Unfair Commercial Practices Directive: a new right to individual remedies for consumers is introduced in Article 11a and the rules on penalties have been strengthened. While the 2005 Directive prohibited a number of practices, it did not introduce specific redress, leaving this issues to the Member States. This is changed in the new proposal. As for penalties, a list of common, non-exhaustive criteria for assessing the gravity of infringements is introduced in Article 13 of the Directive. Particularly significant, however, for ‘widespread infringements’ and ‘widespread infringements with a Union dimension’, Member States will be required to introduce law fines for the maximum amount of at least 4% of the trader’s turnover. In terms of paid placements and paid inclusion, the proposal asks that search results in response to the consumer’s online search query also be included as unfair commercial practice when such paid inclusion is not declared.
  3. 2011 Consumer Rights Directive: the proposal has a number of important additions. It brings the Directive in line with the ‘digital content’ and ‘digital services’ as defined in the proposed Digital Content Directive. A new Article 6a provides specific additional pre- contractual information requirements for contracts concluded on online marketplaces (including ranking parameters, whether the party is a trader, consumer rights arising, etc.), thus increasing consumer transparency in online marketplaces. The proposal also amends the Directive in terms of a number of other issues between traders and consumers, effectively removing unnecessary burden from businesses.
  4. 1993 Unfair Contract Terms: a new article on penalties in inserted.
  5. 1998 Price Indication Directive: article on penalties is amended.

While it is difficult to give a comprehensive assessment of the changes at this early stage, it seems that the new provisions on collective redress and significantly increased fines would have an immediate effect. EU consumer laws are already among the most robust in the world and this reform would both update them and strengthen them.3 On the other hand, it also seems clear that more frequent and increasingly invasive privacy incursions, coupled with ever more subtle modes of tracking consumers and influencing their behaviour, warrant new approaches. It is, therefore, questionable to what extent the EU framework is capable of answering these challenges.

  1. The 1993 Unfair Contract Terms Directive, the 2005 Unfair Commercial Practices Directive and the 2011 Consumer Rights Directive are the key laws.
  2. In that no open calls for its thorough reform have been heard.
  3. This is in addition to new privacy rules, coming into effect in May 2018.

Rule by Decree Part II: How the Commission Undermines Rule of Law by Attempting to Regulate Online Content by Issuing Edicts

In its most simple form, the rule of law can be defined as a way of restricting arbitrary power by subjecting everyone to democratically passed and fairly applied and enforced laws. Such laws are subject to proper procedure and judicial oversight. Rule by Decree, on the contrary, is represented by quick and unaccountable creation by laws favoured by despots.

In March 2017, I have commented on this blog, on the Commission’s attempt to regulate large platforms such as Facebook, Google and others by politely asking them to take action in the field of unfair contracts terms and removing fraud & scams. I have commented at that point that, rather than giving the Commission ex ante enforcement powers, the Treaty only gives it the right to propose laws and to occasionally partake in oversight of the ones that already exist. Where such laws do not exist, because the proper agreement lacks, they cannot and should not be replaced by letters demanding action from individual corporations.

In September 2017, the Commission extended its ambition to regulate platforms by proposing to monitor illegal content on online platforms. In that dubious document, the Commission asserted that “what is illegal offline is also illegal online” and demanded that “platforms” step up the fight against illegal content online. In summary, the document demanded that platform do more to tackle illegal content and take “proactive measures” to detect and remove such content. The Commission kept repeating that the proposed demands do not conflict with the limitations of ISP liability of Articles 12-15 of the E-Commerce Directive but did little to explain how this conflict could be resolved in practice.1 Further to that, different kinds of “illegal” content were all bundled under one title thus putting such issues as child pornography, hate speech, terrorism, commercial fraud and copyright infringement all under one hat. Already at that point, the Commission had been warned that its position is contradictory and dangerous for free speech.

Good laws cannot be replaced by decrees.

Having learned little from the criticism, the Commission marched on and issued a set of operational measures (in the form of Recommendation) leaning on the September 2017 Communication and targeting “companies and Member States” and threatening legislation if these fail to be effective. In it, the Commission calls for

  • Clearer ‘notice and action’ procedures
  • More efficient tools and proactive technologies
  • Stronger safeguards to ensure fundamental rights, in particular when automated tools are used for removal
  • Special attention to small companies
  • Closer cooperation with authorities

Special rules are introduced for “terrorist” content online and this inlcudes

  • the obligation to remove such content within an hour of referral
  • faster detection and removal
  • better referral
  • regular reporting to the Commission, in particular about referrals

While the criticism concerning the September 2017 communication still stands, new concerns have to be added. They can be summarised as follows:

  1. The Guidelines lack effective safeguards against abuse. Points 1.11 and 1.12 require only that content providers be given the right to contest the decision “within a reasonable time period“ and through a “user-friendly” mechanism. Point 18 encourages “proportionate and specific proactive measures” in tackling illegal content in particular by automated means. Points 19 and 20 call for “effective and appropriate safeguards” but, essentially, leave this issue to Member States, further increasing disparity between them and fostering confusion. Point 21 unhelpfully and vaguely calls for protection against abuse without being specific on what this protection might constitute in.
  2. The Guidelines are in direct conflict with Article 15 of the E-Commerce Directive. That article explicitly states that Member States shall not impose a general obligation to monitor the information which they transmit or store, nor a general obligation actively to seek facts or circumstances indicating illegal activity. While it is perfectly possible and allowed, in a civil and democratic society, to repeal this article, this has not happened yet.
  3. The Guidelines bundle different kinds of illegal content. Illegal content is defined in point 1.4.b as “any information which is not in compliance with Union law or the law of a Member State concerned”. The definition is unacceptably wide. Furthermore, it conflates content which is illegal at EU level (because of harmonisation efforts) with material that is illegal only in some states which, in turn, also brings subsidiarity into play (for, why would Member States transfer issues concerning locally-illegal content onto EU). Further to that, copyright, hate speech, child pornography and terrorist content require vastly different monitoring and enforcement tools. It makes no sense to treat them as one.
  4. Finally, and most importantly, good laws cannot be replaced by decrees. The Commission understands all too well that an effort to pass legislation on hate speech alone (never mind anything else covered by the Recommendation) would require lengthy public consultations, acrimonious disputes between stakeholders and endless negotiations as texts progress through various committees in the effort to find a compromise between the Parliament and the Council. Whereas soft law is a legitimate and useful governance tool, it is defined by its non-binding character.2 A recommendation that threatens legislation unless it reaches its desired effect is not soft law, it is a decree. We should remind the Commission that the reason why EU Internet laws have been a success3 is precisely because democratic procedures have been respected. It may be tempting to make shortcuts but this undermines the legitimacy of EU institutions and will ultimately do little to combat illegal content.
  1. In fact, it got into trouble for exactly the same reasons with another badly-formulated proposal – Copyright in the Digital Single Market.
  2. As evidenced, among others, in Article 288 TFEU.
  3. The E-Commerce and InfoSoc directives have lasted over 17 years while the Data Protection Directive is over 20 years old.

A Short Explanation of the Uber Judgment: How to Regulate “Composite” Services

On December 20, the CJEU delivered its long-anticipated judgment in the Uber case. This follows the application, submitted on 16 October 2015, and Advocate General Szpunar’s opinion delivered on 11 May 2017. The anticipation which surrounds the case arises from the potential it may have on disruptive business models that rely on electronic commerce.

The dispute arises from an action before Spanish courts by a Barcelona taxi association seeking a declaration that Uber Spain violates Spanish competition laws. In order to answer that question, it was necessary to determine whether Uber needed prior administrative authorisation which, in turn, required that it be determined whether Uber services were transport services, information society services or a combination of both. It is the latter question that got referred to CJEU. If Uber is an information society service (which was Uber’s argument in the main proceedings), it would not require authorisation. If, on the other hand, it is a transport service (the taxi association’s argument), it would be subject to the same authorisation as regular taxi associations.

Both the AG’s Opinion and the judgment take the view that services which Uber provides are a combination of transport and electronic commerce, and the bulk of the subsequent analysis concentrates on determining how that affects Uber’s services.

Advocate General Szpunar points out in his opinion (paragraph 64) that “within the context of this service, it is undoubtedly the supply of transport which is the main supply and which gives the service economic meaning”. Furthermore, he says that the connection part of the service is merely preparatory and meant to enable the transport part (p. 65). Crucially, since it is “neither self-standing, nor the main supply, in relation to the supply of transport”, it cannot be considered to be an information society service (ISS), within the meaning of the E-Commerce Directive. Paragraph 87 of the Opinion suggests that E-Commerce Directive is explicit in only covering information society service activities while not suggesting that a classification of an activity as being an ISS can, in and of itself, render other parts of the service meaningless. The Directive, in other words, says nothing about whether transport or any other laws also apply to the service. The central issue, the AG suggests, is whether the provider of the service exerts control over the key conditions governing the supply of transport. Where this is not the case, there may be talk of a service being an ISS. But, where this is the case, the connection part of the service does not turn a transport service into an ISS one. In AG’s view, the e-commerce component cannot stamp its spirit on the non-electronic part it depends on.

The Court’s final judgment follows the spirit of AG Szpunar’s opinion, although it uses a somewhat different reasoning. It notes (paragraph 32) that transport and intermediation are different services:

In that regard, it should be noted that an intermediation service consisting of connecting a non-professional driver using his or her own vehicle with a person who wishes to make an urban journey is, in principle, a separate service from a transport service consisting of the physical act of moving persons or goods from one place to another by means of a vehicle. It should be added that each of those services, taken separately, can be linked to different directives or provisions of the FEU Treaty on the freedom to provide services, as contemplated by the referring court. (emphasis added)

The Court is essentially saying that two components of a service may each be subject to different rules. While intermediation is an ISS, “non-public urban transport” is a transport service. Having that in mind, however, the Court concludes (p. 40) that “intermediation service must thus be regarded as forming an integral part of an overall service whose main component is a transport service and, accordingly, must be classified not as ‘an information society service”. The intermediation service too is a service in the field of transport. Similar to AG, the Court thinks that it is the dominant transport component which subsumes the electronic intermediation component.

Although the Court does not spell it out, the main criterion must be the separability of the two services. This is also apparent in the Court’s press release on the judgment, which concentrates on whether the services are “inherently linked”. In the Court view, services which are not inherently linked may be treated separately. If services are inherently linked, i.e. it is not possible for them to be supplied separately, it is necessary to look for the economically dominant one. In Uber’s case, the whole point of the service is for the electronic connection to facilitate the ultimate goal – transport. The electronic component would be meaningless on its own. This echoes paragraph 31 of AG’s opinion:

It would be pointless only to liberalise a secondary aspect of a composite supply if that supply could not be freely made on account of rules falling outside the scope of the provisions of Directive 2000/31.

The service in point, Uber, would have no self-standing economic value without the transport component which, in turn, is regulated elsewhere. The answer must therefore be that composite supply services depend on all relevant legislation and not only on the electronic commerce laws. The fact that they are also electronic commerce services does not take them outside the scope of transport laws.

The judgment offers guidance on what is to be considered a composite service:

In the case of composite services, namely services comprising electronic and non-electronic elements, a service may be regarded as entirely transmitted by electronic means, in the first place, when the supply which is not made by electronic means is economically independent of the service which is provided by that means.

For a service to fall entirely into the ISS category, the non-electronic component must be independent of the electronic one. The present framework leaves the determination of how non-electronic components of composite services are to be regulated to national laws (p. 47 of the judgment) in all cases which are not directly harmonised.

In conclusion, the Uber case recognises several categories of services:

  • non-composite electronic commerce services (e.g. taxi directories), subject to E-Commerce laws only
  • non-composite transport services (e.g. regular taxi services), subject to transport laws only
  • composite services with economically dependent components (e.g. Uber), subject to transport laws (or other dominant non-electronic component)
  • composite services with economically independent non-electronic component (e.g. airline ticket portals), which are subject to E-Commerce laws

It is difficult to disagree with the Court’s logic. This is for a simple reason: a service whose main non-electronic component is otherwise regulated, should not be able to escape such regulation by introducing an e-commerce aspect. To hold otherwise would mean that medical, financial or tourism services would be able to circumvent regulation by having an innovative electronic component. Contrary to the view that CJEU misunderstands the innovative and disruptive aspects of Uber, the Court simply believes that the innovative component in the service may be located in either of the two parts of the composite service. In Uber’s case, this is both in the electronic and the transport components.