Researchers Claim Facebook’s Ad Library Is Deeply Flawed And Can’t Reliably Identify Political Ads

The US midterm elections are coming near and that’s why the digital world is pulling up its gear, making plenty of changes to help tackle the issues that come with it.

We’ve seen so many apps such as Twitter and Facebook make promises to help curb the spread of misinformation. As it is, the issue has turned into a global concern, especially when it comes to US politics.

But now, another major concern is growing after a set of researchers were able to outline how the app lacks the inert ability to correctly and reliably distinguish political-themed ads from all others.

Meta was seen taking immense pride in its Ad Library but it now appears that it may be biting the dust or perhaps speaking too soon, as this new report reveals.

For those who aren't aware, the Ad Library is a tool used by the app that puts transparency as the highest priority. It enables general members of the public to come forward and witness how ads were put and who they were put by. Also, they unveil who is behind the funding.

This sort of data is commonly taken advantage of by researchers and gets cited for other firms to follow as an example.

Recently, a report was published by researchers hailing from NYU and even from KU Leaven. And it boldly found how the ad library belonging to Facebook is greatly flawing, meaning it lacks the ability to reliably figure out which ad is political and which isn’t.

There were some worrying facts that stated how the app actually tends on missing out on more ads than it detects. Similarly, the findings spoke about the platform using incorrect labels to categorize ads that weren’t even labeled by the advertisers in the first place.

The fact that more than 50% of ads detected are wrongly flagged is definitely a major eye-opener for obvious reasons.

But Meta has argued and disagreed greatly with this thought. They claim that the research report had its own shortcomings. In the same way, it says that the authors failed to analyze all the ads that the app blocks before they’re allowed to come forward and run.

And in case you didn’t know, the company doesn’t give anyone, be it researchers or the public, access to such data.

Right now, all eyes are on November of this year, as that’s the date set for the midterms. Now, it’s going to be awfully interesting to see what happens as the last time, armed mobs stormed the US Capitol building. There was a desire to overthrow the presidential elections.

At that time, Meta and Twitter faced intense criticism for failing to react and force their advertising policies to come into play and begin political violence.

Remember, even today, Meta is one of the few who doesn’t ban political ads. In fact, it sells them. The only time it places a ban on sales is just one week prior to the election date.

The paper took ads from July 2020 to February 2021 into consideration and that’s where they found such alarming findings. Out of a sample size of nearly 189,000 ads, around 83% of those labeled to be political weren’t. And that’s a huge error on that app’s part.

There are certainly plenty of questions raised in people’s minds. Meta has not had the best working relationship with researchers from NYU. This is partly related to the way they use scrapers to gain data from the app’s Ad Library.

And in case you weren’t aware, Meta forbids such acts. They even announced to take legal action against such reports in 2020, before the elections.

In this recent paper, NYU clarifies that the data taken didn’t affect any non-ad Facebook user.

H/T: Forbes.

Read next: Meta In The Hot Seat Again As Its Decision To Introduce End-End Encryption Faces Great Opposition
Previous Post Next Post