(Brussels, March 28, 2023) – A European Union regulation under consideration will improve public scrutiny of paid political messaging and limit the circulation of political ads based on people’s personal characteristics and sensitivities, Human Rights Watch said today. But changes are needed to strengthen its protections and to prevent capturing unpaid political discourse.
The European Parliament, Commission, and Council are negotiating the proposed regulation, which would place limits on political advertising that intrudes on people’s privacy while also requiring more transparency in paid political messaging on digital platforms and in other media. But the regulation does not go far enough in its safeguards while risking collateral harm if applied too broadly to civil society groups and journalists participating in political discourse online.
“Targeted online political ads can threaten privacy and undermine the integrity of political processes,” said Frederike Kaltheuner, director for technology and human rights at Human Rights Watch. “Measures are needed to curb practices that intrude on people’s privacy while shining a light on the murky ways in which paid political messages circulate.”
The risks associated with paid political advertising are well documented. By creating a targeted messaging ecosystem, platforms allow political actors to tailor misleading or discriminatory messages based on people’s intimate personal characteristics. The voter profiles that fuel this targeted messaging are also growing increasingly intrusive and detailed, with one political party boasting to have amassed over 3,000 data points on each voter in 2020 while political data brokers include inferences on voters’ credit ratings and political views in voter profiles for sale.
Platforms also rely on algorithmic systems to make decisions about how promoted content is targeted, and how far a paid political message will spread. It is not always clear precisely how these algorithmic systems select target audiences for advertisements and circulation can be indirectly intrusive or discriminatory based on these algorithms.
An industry of online influencers, reputation management companies, and political data brokers also market voter profiling services to take advantage of the online messaging environment. Ties between this industry and political parties are often invisible, further obscuring the sources of political messaging in an atmosphere where hateful, woefully misinformed, or deeply polarizing messaging too often prevails.
These combined factors make it difficult for individuals to understand why they are seeing a specific advertisement and create stiff barriers when regulators, journalists, and researchers attempt to subject online political discourse to public scrutiny.
A study of the 2019 Indian general election, for example, used transparency tools available at the time to show that political advertisements were predominantly sent to men. But without more effective tools, the study could not determine whether political parties actively chose to discriminate in circulating their advertisements or if this was an indirect function of the platform’s audience selection mechanisms.
While platforms have taken some voluntary steps to improve privacy and transparency in political messaging, not enough has been done and even current good industry practices are often unevenly applied and subject to change.
A 2021 study from Privacy International, for example, found that platforms were inconsistent in extending political advertisement transparency measures to different regions. And a 2022 study from Global Witness showed that a voluntary ban on misinformation in political advertisements was far less successfully applied in some jurisdictions than others, while in the 2022 Brazilian presidential election, paid political advertisements falsely challenged the integrity of outgoing President Jair Bolsonaro’s electoral loss.
Against this backdrop, the proposed regulation would require platforms to provide more detailed and standardized archival documentation of paid political messages for researchers to study as well as real-time explanations to users regarding why they were chosen to receive a particular message.
“Improved transparency alone will not solve problems inherent in digital political advertising,” Kaltheuner said. “But it can help document problems by allowing independent observers to better assess whether advertisements are being circulated in a discriminatory or manipulative manner and study any resulting impact on voting.”
Greater public scrutiny is particularly important as paid misleading content continues to spread on platforms despite voluntary efforts to ban its circulation.
The regulation would also require consistent application of platform transparency measures across all forms of political advertising. Current voluntary practices often exclude branded messaging from sources such as online influencers or reputation management companies paid by political parties.
But some of the regulation’s proposed transparency obligations do not apply to smaller platforms, where 10 percent of political advertising is estimated to occur. This exemption could push bad actors to these unregulated platforms in an effort to avoid scrutiny. And by limiting its scope to paid messages directed at EU political events, the draft regulation leaves diaspora communities within the EU vulnerable to misleading or hateful ads about foreign elections on platforms subject to EU jurisdiction.
The regulation also provides additional safeguards for the use of sensitive special category data such as political or religious beliefs to target political messaging, but an European Parliament position currently being negotiated would offer additional critical protections. If adopted, it would place strict limits on the use of all personal data for political advertising while targeting based on special category data or personal data inferred from people’s online actions would be largely banned altogether.
Sensitive personal characteristics such as ethnicity, religious beliefs, political views, and sexual orientation should not determine what paid campaign messaging people receive, Human Rights Watch said. The European Parliament’s proposal will provide critical privacy protections and limit discriminatory targeting.
While it is important to capture all paid political messaging, there is also a risk that the regulation would inadvertently apply to online content creators, civil society groups, and journalists simply for participating in political debates on social media platforms. The regulation should be amended to clarify that its obligations only apply to paid political advertisements and paid political content promotion. This should include, for instance, influencers who are paid by political parties to promote political content.
Finally, the regulation should encourage coordination or overlapping responsibility over privacy issues for independent election bodies and data protection authorities in EU member states. Absent this type of coordination, many campaign-related data protection complaints risk being left uninvestigated and privacy concerns may not be adequately considered by election oversight bodies.
EU institutions should work together to improve the safeguards in the regulation and move to enact together with member states in a timely manner.
Relying on sensitive personal characteristics to choose political ad recipients intrudes on their privacy, and the ways platforms opaquely select audiences can defy attempts to scrutinize often-divisive paid political messaging, Human Rights Watch said. The draft law is helpful in addressing these problems but should embrace European Parliament’s proposed safeguards and any ambiguity that it might apply to unpaid political messages should be erased.