Facebook app on the smartphone: The social network is an important factor in the Brazilian election campaign
Photo: Jenny Kane/AP
The mood is heating up ahead of the presidential election in Brazil in early October, but the Facebook group Meta is failing in the fight against disinformation, claims the NGO Global Witness.
In one experiment, the experts deliberately violated Facebook's own rules for the Brazilian election campaign, but their ads were still released.
Meta itself is currently promoting its intensified efforts against disinformation.
As the group emphasizes, it now employs more than 40,000 people worldwide in the fight against misinformation and targeted campaigns, and invests five billion dollars.
In the run-up to the Brazilian elections, it is working with the relevant authorities to educate voters on how to vote and how to identify misinformation.
The group also prevents false reports from spreading unchecked on WhatsApp.
Experiment with improper advertising
"For the first time, the Supreme Electoral Court will be able to report content directly on Facebook and Instagram that may violate our guidelines," Meta describes its own efforts.
The company told SPIEGEL that before the local elections in 2020 it had deleted a total of 140,000 posts on Facebook and Instagram and rejected 250,000 illegal political ads.
The organization “Global Witness”, based in London, Brussels and Washington, has put the current efforts to the test: They placed ten ads that clearly violate the rules Facebook set itself for the Brazilian election campaign.
Five of the ads spread false information about the election process, such as where and when to vote.
Five other ads contained messages delegitimizing the election itself.
This includes, for example, the rejection of voting computers, which is currently being fueled by the incumbent Jair Bolsonaro himself.
The result: All ten ads were approved by Facebook.
However, since the ads were not played out, some of the measures taken by the social media group were not effective.
They are based on users reporting suspicious posts.
However, Meta claims that its own algorithms can recognize manipulations without user input.
Thanks to artificial intelligence, they could remove 99.7 percent of fake profiles before users reported them.
According to the company, the non-governmental organization made it particularly easy for the group to discover the manipulations.
She didn't place the ads from Brazil itself, but from IP addresses in Nairobi and London - actually clear alarm signals for the moderators who use Facebook.
"It is clear that Facebook's election integrity measures are simply ineffective," writes Global Witness in its conclusion.
"Meta must recognize that protecting democracy is not optional: it is one of the costs of doing business." The non-governmental organization is therefore calling for the group to intensify its efforts in this and all other election campaigns.
This is intended to provide better training for moderators and make all ads placed transparent for third-party investigation.
It is true that Facebook publishes election advertising in a transparent process that discloses the originator of an ad and the expenses.
But to do this, the advertisements must first be marked as election advertising by the clients themselves.