Hosting Platforms after the Italian GoogleVideo Case – Data Controllers or not?

In its long awaited judgement, the Italian Supreme Court ruled that Google Video could be not be deemed a data controller with regard to the videos it hosts on its platform. As a result, they cannot be held responsible for the dissemination of these videos. The Court specified that the rules ‘presuppose actual decision-making power over (a) the purposes and means of the relevant processing (dissemination to the public); and (b) the balancing between different rights and interests at stake. It can be deduced from the existing framework that this decision making power depends on the existence of actual knowledge. In other words, Google Video only becomes responsible (data controller) from the moment it is made aware. This interpretation, the Court explained, is in line with what is written down in the eCommerce Directive (exemption of hosting providers and no general obligation to monitor, artt.14-15).

It is worth saying, however, that many processing activities are relevant in this context. The dissemination of the video (containing personal data) is one processing activity, for which the uploader should be considered controller. But, besides this, the video (and hence the personal data contained within) is potentially subject to many other processing activities as well (analysis for behavioural marketing purposes, facial recognition, etc.). With regard to this second strand of uses of the data, a strong argument can be made for the hosting platform to be the data controller. After all, they are determining the purpose and means of these specific activities.

Because in the case at hand, it was mainly the activity of dissemination that was objected to, the original controller bears primary responsibility. But this should not overshadow the responsibilities of hosting platforms (and the like) for the plethora of other processing activities the data is subject to.


Intermediary Liability – Automation = Neutrality = Exempted?

Tomorrow, Advocate General Jääskinen is to release his opinion in the much debated Google v Spain (aka Right to be Forgotten) case. According to Google’s adversaries, Search Engines are to be considered data controllers under the Data Protection Directive for the personal data (on the websites) they refer to and are therefore (under certain circumstances) liable to remove links (for more info, see my previous blogpost on this case).

An often invoked counter-argument to liability assertions by Internet intermediaries relates to the automated nature of their processing activities. In other words, Intermediaries often argue that they are merely providing a neutral service, content-agnostic and fully automated. After all, it is claimed, decisions are made by algorithms and no human eyes actually ‘see’ the information. In 2010 the CJEU seems to have acknowledged such an interpretation, stating that services that only perform activities ‘of a mere technical, automatic and passive nature … having neither knowledge of nor control over the information which is transmitter or stored’ should be exempted (Louis Vuitton v Google, C-236/08). In the UK, Justice Eady has also ruled that automation precludes intentionality (In this case, Google was not held liable for defamatory snippets it displayed in its search results.). The argument that automation equals neutrality, however, seems to be falling apart. Being a mere (automated) ‘organizing agent’ does not necessarily entail neutrality, nor does it necessarily validate the exemption of liability. After all, as U. Kohl aptly describes: “both organization and automation require human judgment and thus have assumptions, values and goals embedded into them.” Put differently, the use of algorithms does not imply neutrality. Instead of looking at the automated nature of an intermediary service provider, one should look at how it is designed. Such interpretation was also followed by AG Jääskinen in L’Oreal v eBay. In his Opinion, he clearly stated that neutrality does not even constitute the right test to decide on the exemption from liability. Instead, one should look at the type of activities of the relevant service provider. The liability of a search engine, for example, may depend on the fact whether it simply refers to information, displays snippets or autosuggests the information. Particularly with regard to defamatory content, a considerable amount of – widely diverging – cases has emerged over the last few years. A number of French Courts have ruled search engines to be liable for auto-suggesting information. Very recently, the German Federal Court of Justice also overturned two decisions, requiring Google to block defamatory auto-complete search suggestions. In a 2011 Italian judgment, the Court followed the plaintiff’s claim that Google should be considered to have ‘created’ the auto-complete terms. Even if not considered to actually create the actual terms, search engines still make a deliberate decision to adopt autocomplete functionality and design the underlying algorithms. Moreover, it is particularly hard to rely on the neutrality argument (based on automation and alleged content-agnosticism) after the intermediary has received a notification (infra).

Put briefly, we see a crumbling of the ‘automation=neutrality=exemption-from-liability’ argument. Particularly taking into account AG Jääskinen’s previous statements in L’Oreal v Google (specifically expressing concern on the argument that “in order to establish whether the liability of a paid internet referencing service provider may be limited under Article 14 of Directive 2000/31, it is necessary to examine whether the role played by that service provider is neutral in the sense that its conduct is merely technical, automatic and passive, pointing to a lack of knowledge or control of the data which it stores”), it is most likely that he will come to a similar conclusion in his Opinion in the Google v Spain Case which will – normally – be released tomorrow.

Google v Spain at the Court of Justice of the EU

One week ago, the so-called ‘Google v Spain’ or ‘Right to be Forgotten’ case was heard before the Court of Justice of the EU (C-131/12).

Put briefly, Google was ordered by a Spanish Court to remove certain search results – relating to Mr. Carlos José – from its index. The contentious search results linked back to Mr. José’s insolvency proceedings, published in a newspaper in 1998. Google appealed and the Audiencia Nacional referred to the ECJ, lodging for a preliminary ruling on the three main questions in this case: (1) Territoriality: do the contentious facts fall within the territorial scope of application (art.4) of the Data Protection Directive (DPD)? (2) Can Google – as a Search Engine (SE) – be considered a (personal) data controller? (3) What is the extent of the ‘Right to be Forgotten’ aka ‘right to erasure and blocking of data’? Currently there are over 200 similar cases pending before Spanish Courts (and not unlikely in other EU jurisdictions as well).



(1) Territoriality

▪       The contentious activity at stake here is not ‘carried out in the context of’ the activities of Google Spain – being an EU establishment of Google Inc. (Article 4(1)(a) DPD). Google Spain is not involved in the SE activities itself (Google Inc is). Its only relevant activities with regard to the SE is to provide advertising space and their behavioural advertisement model is not based on the indexed content.

▪       Art.4(1)(c) is not applicable either, as the mere provision of a service in an EU Member State (even if Spanish domain name is used) cannot be considered ‘use of equipment’ within the meaning of the DPD. The use of web spiders to index content should not be considered ‘use of equipment’ either. The use of cookies would constitute ‘use of equipment’ but is not relevant in this case.

(2) Google as Controller

▪       Google collects, processes and indexes data indiscriminately. They are ignorant about whether or not the content of the webpages that is being indexed contains personal data. There is an obvious lack of intent that distinguishes this case with the Lindqvist and Satamedia cases.

▪       The decision to take content down should be taken by the party that is best placed to do so. Given the fact that Google does not control any of the data held on the websites it indexes, nor that it has any intent to do so, the publisher of the original webpage is best placed to decide.

▪       Even if one would consider Google to be processing personal data, the company still argues not to be a data controller because: (a) there is no intent to process personal data; (b) Google does not verify whether indexed data is personal or not; (c) the publisher has final and continuing control; (d) if the publisher removes the (personal) data, Google does so as well; (e) Google cannot verify the legitimacy personal data processing; (f) Google only plays a secondary/accessory role in disseminating the information; (g) articles 12-15 eCommerce Directive; (g) Article 29 Working Party’s Opinion 1/2008 (p13-14) endorses SE’s role as intermediaries in situations such as the one at hand; (h) Google’s role can be compared to that of a Telecom operator who are stricto sensu  also processing personal data, but or not liable under the DPD either. Mere intermediaries transferring data.

(3) Right to be Forgotten/Erasure v-a-v Search Engines

▪       On the question whether a SE can be asked to remove links directly, without the data subject first having to go to the original publisher first, Google raises the Freedom of Expression (FoE) and Freedom of Information (FoI) flag: (1) the original publishers will be deprived of an important channel of communication and (2) Internet users in general will have access to less information. The responsibility of publishers should not be shifted onto the shoulders of intermediaries such as search engines. This would also violate the proportionality principle, Google argues: (a) Obliging Google to erase search results does not prevent the information form appearing elsewhere; (b) Google cannot assess the legality; (c) Google can only remove the link to the webpage entirely (it cannot just anonymise certain bits in the webpage), which would constitute overkill as the webpages will usually contain much more information than just the contentious personal data.

Plaintiff (+ allies: Spanish & Austrian Government, European Commission)

(1) Territoriality

▪       For the DPD not to apply to the issue at stake, the question is whether or not Google Spain’s activities can be sufficiently distinguished from those of Google Inc. This is clearly not the case, according to the plaintiff (and its allies). Google Spain’s activity is not merely ancillary, but constitutes an integral part of Google Inc’s activities. They are just doing it for a particular jurisdiction.

 (2) Google as Controller

▪       DPD was written before the widespread use of the Internet and SE’s in particular. The DPD should, therefore, be applied creatively to Google. According to the plaintiff, Google is a data controller as it actively collects and processes personal data (referring to art.2(b) DPD ‘dissemination or otherwise making available’ and Lindqvist: any processing of data – even if data is published already – constitutes PD processing within scope of DPD). Its activity constitutes a separate ‘publication’ from the original one.

▪       Google can even be considered a data controller v-a-v the content of the webpages it indexes, because: (a) it determines the means (algorithms, web spiders, etc.) and purpose (include information in search results) of processing; (b) Google actively directs and controls the flow of information and its actions cannot be compared to bidirectional traffic management of a telecom operator. In other words, they are not ‘neutral’ intermediaries. Google provides an added value service, which it cannot provide without acting autonomously.

▪       It was also argued that the criteria of art2(a) and (b) are ‘objective’ in nature. The intent of the ‘controller’ is not relevant. Hence SE’s are data controllers as they are de facto processing personal data.

▪       Google – allegedly – also has a separate responsibility because it makes information much more widely available, it can provide a complete picture of an individual and it has its own specific (commercial) purposes. Google does not “control” the initial uploading of content, but it does control its aggregation and subsequent dissemination. Responsibility of SE’s is distinct and separate.

(3) Right to be Forgotten/Erasure v-a-v Search Engines

▪       It is stressed that Google is not asked to conduct a priori monitoring of all the content it indexes. The plaintiff (and its allies) rather advocate for a specific notice and takedown regime, similar to copyright claims. Only when a specific complaint is made, regarding a specific piece of content, Google should remove the search results.

▪       In order to invoke such a right, the data subject should at least demonstrate a violation of the legitimate processing requirement (art.7 DPD) or the principle of data quality (art.6 DPD).

▪       On the risk to FoI and FoE: When there is a conflict between different rights, a balance should be made. Neither one should automatically prevail.

After this first round of arguments, the court asked several questions to all parties involved. Most of them related to the practical implications of a potential obligation on Google to remove search results.

The Advocate General will finish his Opinion by June 25th, and a final judgement should follow soon thereafter.

Together with some colleagues at ICRI, I will soon publish a working paper making a more thorough legal analysis of all the issues at stake in this particular case. To be Continued…

Google faced with a million requests a month to remove copyright searches

Google is receiving more than a million requests a month from copyright owners seeking to pull their content from the companys search results, the web giant has revealed. The number requests has grown so fast that it now often tops 250,000 a week, more than Google received for all of 2009.


ECJ rules in favor of Net Freedom and against censorship

Judgment in Case C-360/10 Belgische Vereniging van Auteurs, Componisten en Uitgevers (SABAM) v Netlog NV: 

The owner of an online social network cannot be obliged to install a general filtering system, covering all its users, in order to prevent the unlawful use of musical and audio-visual work

Such an obligation would not be respecting the prohibition to impose on that provider a general obligation to monitor nor the requirement that a fair balance be struck between the protection of copyright, on the one hand, and the freedom to conduct business, the right to protection of personal data and the freedom to receive or impart information, on the other.

Dehli High Court threatens to shut down social networks ‘like China’…

“The Indian government has sanctioned the prosecution of executives from companies like Google and Facebook for objectionable content posted online. Yesterday, the Delhi High Court warned that like China, India can shut down these sites. 21 internet companies are in legal trouble. Charges against them will include “promoting enemity between groups”  and “deliberate malicious acts intended to outrage.” Since some of the charges requested in court are non-bailable, the government had to indicate whether it agreed with the need for the websites to stand trial.”

See: Google, Facebook case: Govt sanctions prosecution over objectionable content.

Australian High Court weighs ISP responsibilities

“The legal issue before the five-judge panel and, indeed in previous Federal Court cases*, is whether an ISP’s failure to take any steps to stop infringing conduct, means it authorised the infringements and thus breached the Copyright Act.

“Justice Emmett ruled that, in order for iiNet to respond to its notices, AFACT had to offer “clear and unequivocal evidence” of infringements, to indemnify iiNet and reimburse it for reasonable costs.”

Read more: iiTrial: High Court weighs ISP responsibilities – Telco/ISP – Technology – News –