Political ads ahead of the 2020 presidential election collected personal data, disseminated misleading information

Online advertisements are frequently splashed on news websites. By clicking on these banners or links, the news site generates income. But these ads also often use manipulation techniques, the researchers say.

Researchers at the University of Washington were curious about what types of political ads people saw in the 2020 presidential election. The team looked at over a million ads on nearly 750 sites in information between September 2020 and January 2021. Among these announcements, nearly 56,000 had political content.

Political ads used several tactics that worried researchers, including masquerading as a poll to collect personal information about people or have headlines that could affect internet users’ opinions of candidates.

The researchers presented these results on November 3 at the ACM Internet Measurement Conference 2021.

“The election is a time when people are given a lot of information, and we hope they process it to make informed decisions towards the democratic process. These ads are part of the information ecosystem that reaches people , so problematic ads could be particularly dangerous during election season, â€said senior author Franziska Roesner, UW associate professor at the Paul G. Allen School of Computer Science & Engineering.

The team wondered whether or how the ads would take advantage of the political climate to harness people’s emotions and get people to click.

“We were in a good position to study this phenomenon because of our previous research on misleading information and manipulation techniques in online advertisements,†said Tadayoshi Kohno, UW professor at the Allen School. “Six weeks before the election, we said, ‘There are going to be some interesting commercials, and we have the infrastructure to capture them. Let’s go get them. This is a unique and historic opportunity.’â€

The researchers created a list of informative websites spanning the political spectrum, then used a web crawler to visit each site each day. The crawler scrolled through the sites and took screenshots of each ad before clicking on the ad to collect the landing page URL and content.

The team wanted to make sure they got a wide range of ads because someone based at UW might see a different set of ads than someone in another location.

“We know that political ads are location-targeted. For example, ads for Washington candidates will only show to viewers browsing from Washington state. Or maybe a presidential campaign will have more ads. in a swing state, â€the lead author said. Eric Zeng, UW doctoral student at Allen School.

“We have configured our crawlers to explore from different locations in the United States. Because we didn’t have computers set up across the country, we used a virtual private network to make it look like our crawlers were loading sites from those locations. “

Researchers initially configured the bots to search news sites as if they were based in Miami, Seattle, Salt Lake City and Raleigh, North Carolina. After the election, the team also wanted to capture any announcements related to the Georgia special election and the Arizona recount. So two crawlers began to search as if they were based in Atlanta and Phoenix.

The team continued to explore the sites throughout January 2021 to capture any commercials related to the Capitol Uprising.

Researchers used natural language processing to classify ads as political or non-political. Then the team manually scrolled through political ads to further categorize them, for example by party affiliation, by who paid for the ad, or by what types of tactics the ad used.

“We saw these bogus survey ads that harvested personal information, like email addresses, and tried to target people who wanted to get involved politically. These ads would then use that information to send spam, software. malicious or just general email newsletters, â€said co-author Miranda Wei, UW PhD student at Allen School. “There were so many fake buttons in those ads, asking people to accept or decline, or to vote yes or no. These things are clearly meant to trick you into giving up your personal data.”

Ads that appeared to be polls were more likely to be used by conservative-leaning groups, such as the conservative media and political nonprofits. These ads were also more likely to appear on conservatively leaning websites.

The most popular type of political advertising were click-bait news articles which often featured prominent politicians in sensationalist headlines, but the articles themselves contained little substantive information. The team observed more than 29,000 of these ads, and crawlers often encountered the same ad multiple times. Similar to bogus survey ads, these were also more likely to appear on right-wing sites.

“One example was a headline that read, ‘There’s Something Fishy About Biden’s Speeches,'” said Roesner, who is also co-director of the UW Security and Privacy Research Lab. “I’m afraid these articles may contribute to a body of evidence that people have amassed in their minds. People probably won’t remember later where they saw this information. They probably haven’t even clicked on it, but it does. always shapes their perspective of a candidate. “

Researchers were surprised and relieved, however, to find a lack of advertisements containing explicit misinformation about how and where to vote, or who won the election.

“To their credit, I think the advertising platforms are picking up the wrong information,†Zeng said. “What is happening are ads that exploit the gray areas of content and moderation policies, things that seem deceptive but comply with the letter of the law.”

The world of online advertising is so complicated, researchers say, that it is difficult to determine exactly why or how certain advertisements appear on specific sites or are seen by specific viewers.

“Some ads are showing in certain places because the system has decided that they will be the most lucrative ads in those places,†Roesner said. “It’s not necessarily that someone is sitting there doing this on purpose, but the impact is always the same – the people most vulnerable to certain techniques and certain content are the ones who will see it more.”

To protect computer users from problematic ads, researchers suggest internet users should be careful not to take content at face value, especially if it looks sensational. People can also limit the number of ads they see by getting an ad blocker.

Theo Gregersen, a UW undergraduate student studying computer science is also a co-author of this article.