Towards A More Transparent Regulation for Online Targeted Ads

Mohamed Suliman is a researcher at Northeastern University Civic AI lab with a degree in engineering he is also a regular contributor to Global Voices where he writes about tech issues.

Big tech companies are notoriously secretive about how they operate online advertisements. Often they release only the minimal data required under the threat of public pressure. Worse, researchers are frequently blocked from conducting independent research that could inform the public about how these ads impact society. Currently, these companies voluntarily share and publish only some of the data pertaining to their advertising strategy. Facebook, for instance, publishes data through its Ad Library, which contains a searchable database of the names and frequency of ads, money spent on them, and disclaimers. It also includes downloadable reports and an Application for Programming Interface (API) that helps connect the database with any external source to run analysis. Google does the same through its transparency report, but there is no unified form or policy. Each company is left free to decide what data to share and in what fashion.

Further, social media companies seem uninterested in cooperating with any independent research or sharing data about their ads systems. For example, Facebook blocked researchers from NYC University who were studying political ads on the platform. By contrast, Twitter banned all political advertisements, but also quietly removed its ad transparency center that used to provide data about how they run ads. These incidences point to the fact that we can’t rely on internal regulation to arrive at an understanding of how these tech companies place their ads and what impact they may cause.

In May 2021, Congresswoman Lori Trahan (MA-03) attempted to address this issue by introducing the Social Media DATA transparency Act. The legislation is introduced to Congress but has not yet been debated or approved. The legislation as it stands is a strong start, but there are some shortcomings that should be addressed before it is passed.

The press release for the Act cited many current online advertising practices as justification for its creation. These practices include manipulating older women to sell high interest credit cards, targeting young users with ads for junk food and pill parties, targeting veterans with ads for predatory for-profit colleges, and targeting patients with ads for fraudulent opioid rehabilitation centers. These are serious violations of citizens’ rights, they have impacted many different groups in society, and they hint at how the whole advertising system functions. Aside from these violations, the automated system built by tech companies to run the ads is not always trustworthy and requires scrutiny. Most recently, a study conducted by researchers from New York University and KU Leuven in Belgium found that of 189,000 political ads that ran on Facebook between July-February 2020 Facebook mislabeled 62% of political ads. This shortcoming in the system shows that we cannot rely blindly on ads libraries provided by tech companies and that an outside body should oversee the collection of this information. With proper data sharing, these violations and misclassifications would have been discovered earlier and legally addressed.

The proposed act addresses many important issues by calling for greater transparency about ads and advertisers and establishing a Congressional working group dedicated to online advertising. However, there are four points that need to be improved and included before moving on with the draft.

First: The Act calls for data access only for academic researchers, or those who conduct research in collaboration with any academic institution and for non-commercial purposes. This strict definition leaves out journalists, NGOs, watchdog groups, and even independent citizens that may want to utilize the data to analyze these platforms and warn us of any violations. Data access must not be a privilege solely for those affiliated with academic institutions, particularly as journalists and activists have a long history of sharing leaked data from tech companies to the public.

Second: The act should also state clearly that any released data must be shared under a general public license. As it stands, big tech has complete control over data sharing. Facebook, for instance, requires that users agree with their platform policy before using its Ad Library API. This policy gives the company full power over the data, allowing them to restrict access and content sharing, and most importantly, to amend the policy at any time. Social media platforms should not have such power. Advertising data must be freely available to share and redistribute to anyone.

Third: While the act covers access to many types of data, it overlooks data pertaining to user-generated reports of harmful ads by social media platforms. This data is crucial because it reveals how the platforms truly respond to reports of fake news and fraudulent ads, and if they are taking reasonable measures to address all these violations. In other words, by forcing platforms to share this data, we give them a strong incentive to act righteously to remove any harmful content.

Fourth: To facilitate analysis of the impact of online ads across all social media platforms, the legislation should request that tech companies adhere to unified ads categorization. Facebook, for example, is currently grouping social issues, elections, and politics in one singular category. By contrast, Twitter has a very different categorization for interest targeting that includes 25 categories and 35 sub-categories. If all platforms adopt universal and precise categories dedicated to each issue it will be easier to conduct research about the same issue across platforms.

Because tech companies and social media platforms rely on online advertising for revenue, we can only expect them to continue to maximize the amount of attention given to the ads run on their websites. Given the extent of these platforms’ influence over our collective social lives, public oversight is required. This oversight, in turn, requires access to data about this advertising.

The Social Media DATA transparency Act is a long-awaited and consequential piece of legislation, and because of its potential impact, we must ensure it is as effective as possible. It is a big step towards taming the power of big tech and building a fair and just advertising system that serves the best interests of society. With a few amendments, it will do just that.