On 25 November 2021, the European Commission has published its comprehensive legislative proposal package to reinforce democracy and protect the integrity of elections. The package comprises a new draft proposal on a new regulation on the transparency and targeting of political advertising and three amendment proposals: one of those on European political parties and two on electoral rights of EU citizens. The package had been prescheduled by the European Democracy Action Plan (2.1. and 2.2.) for year 2021.
The focus of this post is the only brand new legislative proposal (the others are amendments): the draft proposal for transparency and targeting of political advertising. The necessity to regulate this field has emerged in the past five years only.
In 2017, revelations shocked the world, telling how two landmark democratic events, the Brexit vote and the US elections (both in 2016) were unfairly influenced through manipulative advertising techniques that combined micro-targeting with disinformation. Some of them deliberately aimed at generating social tension and polarisation. Whereas disinformation has attracted considerable attention by researchers and policy-makers, micro-targeting has been viewed less critically. The proposed regulation was supposed to react to this claim. However, the Commission, striving to avoid overrestriction of freedom of expression, took an extremely cautious stance.
The new rules
The legislative changes provide for robust, cross-actor obligations for transparency – and not more. Targeting and amplification are briefly discussed and the rules provide hardly more than what has already been provided by the GDPR and what was already voluntarily applied by the major social media companies. Transparency is indeed a first step to see clearly in this rapidly developing sphere, with the potential to empower citizens and to deliver information to make further steps, if any. The transparency obligations are not supposed to exercise a disproportionate chilling effect on the speaker, even when they can be evaluated as rather extensive. The rules on targeting merely limit the use of sensitive personal data, but allow even that if the user consented, or is member of the organisation. In an environment where consenting fatigue and various manipulative techniques to acquire consent are ubiquitous, this can hardly be regarded as anything close to effectively protecting users’ rights or the democratic discourse.
In any case, an EU definition of political advertisement is an interesting endeavour, so let’s take a closer view of that first.
How to define political advertisement?
The current information environment blends professional content with organic content online. Advertisements may get shared organically by individual users, whereas memes that were organically created, might become amplified as parts of a political campaign (although Facebook allows promoting of own posts only). Therefore, the definition of political advertising will be a key element of the regulation. Rather than focussing on the factor of intention, like the definition of commercial advertising (“in order to promote the supply of goods or services“), the proposed regulation’s definition grasps the speakers’ role (see below) and the effect: “which is liable to influence the outcome of an election or referendum, a legislative or regulatory process or voting behaviour.”
The speaker, or advertiser is also way more accurately defined than that in the definition of commercial advertising: rather than “in connection with a trade”, political advertising is “by, for or on behalf of a political actor, unless it is of a purely private or a purely commercial nature”, where political actor is also defined quite precisely. Its definition (a-h) includes, beyond what can be regarded as common sense meaning (a-f), also “political campaign organisations with or without legal personality” (g), and “any natural or legal person representing or acting on behalf of” those listed exhaustively, promoting the political objectives of those (h).
The detailed description’s clarity is diminished by not knowing who shall decide whether a person acts “on behalf of” a political party, or an elected official, or an unelected member of government, promoting the political objectives of any of those. For instance, NGOs are likely fall in this category without their intention, as ones that are seen as following the same agenda as an opposition party, when they are critical of governmental policies. This may be particularly worrying in the light of the anti-civil atmosphere of some Member States.
The definition becomes particularly inclusive with the beneficiary and the effect being alternative conditions: whether the message is on behalf of a political actor, or it is liable to influence elections, any of these conditions will qualify it as a political advertisement. The latter criteria may be capable to include organic content of political nature, for example opinion articles of journalists, messages of NGOs, or simply social media influencers’ content. Remuneration is, according to Recitals 27-29 a condition for qualifying as advertisements. However, this is conspicuously missing from the definition section of the draft.
The definition’s action element also reflects the new challenges. Whereas advertising in the EU Directive concerning misleading and comparative advertising is defined as the “making of a representation in any form”, the draft Regulation is more specific: “the preparation, placement, promotion, publication or dissemination, by any means, of a message”. This leads us an important aspect of the draft Regulation: imposing legal obligations on various actors in the entire value chain of advertising.
Obligations to fulfil transparency requirements are distributed across the entire sector of political advertising: providers of advertising services, political actors and publishers (including social media service providers). Providers of political advertising services will need to use their contractual power to acquire the appropriate declaration of their sponsors about whether the ad is a political one, and to include in the contract how their service complies with the relevant provisions of the Regulation (Article 5). Then, they are obliged to transmit a package of required information to the publisher, and to also retain it for five years after the ad’s last publication. This information should include reference to the campaign, the specific services, the value of the countervalue whether financial or other benefits, and the identity and contact information of the sponsor (i.e. the political actor) (Article 6).
The largest set of obligations is put on the publisher. It must ensure that the ad is labelled as a political ad, and that a range of further information are attached to it in such a technical way that those information remain attached to the ad even when it is further disseminated. This way, organic sharing of the paid content would not deprive it from its legally prescribed safeguards.
The attached information should include a range of elements which are partly set out in the text, partly specified in Annex I which is subject to change by the European Commission which is empowered to adopt delegated acts in this respect. The information should include the period during which the ad is published, the aggregated amount of money or benefits that were used as payment for the entire ad campaign (not only publishing but also preparation, placement, promotion and dissemination) and its sources, links to online repositories, information on the notification (complain) mechanism, and the identity of the sponsor and, lastly, the entity ultimately controlling the sponsor (Article 7). It is undoubtedly a great idea to reach behind eventual proxy persons, and to define the real person of political interest who gave the order to publish the political ad, i.e. the entity ultimately controlling it. Unfortunately, it is not entirely clear how will the publisher reveal this information. Still, at least the publisher has a justification and coverage to demand such information and to publish it, in case it managed to acquire it.
Similarly, it is not known how the publisher can ensure the transparency of the “sources of the funds being used”. Its obligation extends only to “make reasonable efforts” to ensure that the information is complete, but, if they find it is not complete, they shall not publish the ad. It is to be seen, how effectively this relative and flexible obligation will contribute to achieving real transparency. The information should also be transmitted if requested by researchers and other actors, including civil society organisations, political actors, journalists and electoral observers, but there are wide possibilities to reject the request or demand a fee.
These obligations are to be interpreted together with Article 30 of the proposal for a Digital Services Act on repositories of advertisements, with the Regulation on political advertising being the lex specialis that would demand more information to be provided, and extends the time period of display from 1 year to 5 years in case of political ads.
Publishers have a number of further obligations: they need to create an electronic mechanism for reporting unlawful political ads; to provide information to national authorities in case of information request; and include the value received in exchange for political advertising services in their annual financial report (Articles 8-10).
Even in absence of scientific evidence, targeting is widely seen as a cost-effective method of persuasion. Targeted ads are seen only by selected users, who are thought to be more susceptible to get affected by the message. This reduces the ad exposure for uninterested users. But political ads have been seen to apply targeting for purposes of manipulation, by selectively transmitting contradicting messages to different audiences. In an ideal world, the prescribed transparency requirements are supposed to reveal, and therefore, prevent such abuse of targeting. Other than this concern for the democratic public discourse, the main objection against the targeting technique is the use of personal data. The draft Regulation addresses this by theoretically ruling out the use of sensitive (special category) data. However, it practically allows using, following the logic of GDPR, or the prohibition of product placement, which, after a firm prohibition, list the cases when it is still allowed. Sensitive data can be processed with the explicit consent of the data subject, based on paragraph 2.(a) of Article 9 GDPR, and for political parties and other non-profit bodies with a political (etc.) aim to target their members, without their specific consent, taking avail of the paragraph 2. (d). Let’s stop for a minute: is there anything specific in this rule of the proposed Regulation in relation to the GDPR? What other condition, listed in paragraph 2. of Article 9 GDPR could potentially justify the use of sensitive data for advertising purposes? Wouldn’t the use of personal data for targeting purposes genuinely be conditional on consent? The only reasonable categories are those mentioned: explicit consent and membership. We must acknowledge that it would be unreasonable to exclude these possibilities. Didn’t we expect something more from a specific law on political advertising? Even Recital 47 of the draft Regulation says that “Additional restrictions and conditions compared to Regulation (EU) 2016/679 and Regulation (EU) 2018/1725 should be provided.” Those additional conditions and restrictions seem to be missing from the substantive part. What could have been added? For example, a prohibition of profiling, whether based on general data, or sensitive data collected with consent. It is known (and mentioned by the Recitals as well) that inferred data is information generated from other, readily available data. General data, such as pictures, or simply the written comments, can be used for deductions on sensitive information, to racial identity. The draft Regulation does not reflect on this possibility. Another possibility could have been to guarantee users influence over the criteria according to which content is made visible for them, as proposed by the Committee on Legal Affairs of the European Parliament.
It sounds cynical from the European Commission to build on the GDPR the new legislation with a greater stake than before, leaving unaddressed the existing legal practice, which systemically violates the GDPR principles on data protection by design, and on on what should constitute “consent”: freely given, specific, informed and unambiguous indication of the data subject’s wishes. It may be understandable that the Commission did not intend to reiterate, or add more details to the expectation on what constitutes “explicit consent”. The detailed expectations, including those of explicit consent, have been already set out in the Guidelines of the EDPB on consent, which, as well as the relevant principles of the GDPR, are simply ignored by the vast majority of service providers. Moreover, in the absence of explicit mention, we must assume that non-sensitive data can be used even if the data subject has not consented. This would be an absurd, but logical conclusion (argumentum a contrario).
In the light of this, the targeting-related transparency requirements seem less than relevant. Even if there is an internal policy in place, retained for five years along with information on the used mechanisms, techniques, including the sources of the data and the parameters of targeting – who will be in charge of reviewing these and along what critera? It looks like NGOs and perhaps, political opponents may go through these data-jungles in search for something morally compromising – as legally, there is not much they can bring up against.
Thus, the proposed Regulation foresees a multi-stakeholder liability by placing obligations on providers of advertising services (notably creative agencies), publishers, data controllers, and political actors – but especially on users and civil society, to keep themselves aware, informed and critical despite all the odds, and follow the small letters in political advertisements and study online repositories.
The Commission opted to specifically refrain from addressing content, in order to protect freedom of expression as much as possible. This is a noble goal, which we could only support, would be not for the fact, that the micro-targeted political ads carry the risk of reducing publicity of rational political discourse. Even the European Court of Human Rights allowed limitation of political ads when it found that the contrary may actually diminish the diversity of opinions in the society. A limitation on political micro-targeting would in fact restrict not the speech, but the silence, by obliging political actors to share their ads with the whole society rather than only with a selected minority. What may look on its face as a limitation of freedom of expression, would ensure more information and transparency, would trigger counterspeech and political discussion on controversial matters. Deliberately limiting the scope of audience exposed to divisive political topics contributes to the already painful division of democratic societies.
Judit Bayer is Senior Research Fellow of the Institute for Information, Telecommunication and Media (ITM) at the University of Münster, and associate professor of the Budapest Business University, Department of Communication.