Advertisements with clearly false messages on October elections were published by the social network, despite many signals which should have triggered an alert, denounces the NGO Global Witness.
Le Monde
The non -governmental organization (NGO) Global Witness made public , Monday, August 15, the results of a testing campaign that she carried out in Brazil to measure the effectiveness of moderation practices of political advertisements in the country.
Brazil will vote on October 2 for the general elections; The election mainly opposes the outgoing president, Jair Bolsonaro (far right), to the former president Luiz Inacio Lula Da Silva (left). As in previous polls in the country, the campaign was marked by major disinformation campaigns, mainly for the benefit of Mr. Bolsonaro.
Global Witness has created and tried to put online several campaigns of political advertising containing clearly false information: bad date for the ballot, misleading messages on correspondence voting … The account that bought these advertisements was located outside Brazil, Advertisements were paid for foreign currencies and the account had not confirmed its identity, which is normally compulsory to be able to publish political advertisements on Facebook. So many alert signals which would have, in theory, had to put the chip in the ear of Facebook moderation services.
According to Global Witness, all these advertisements have been validated and published on Facebook, except one. And, at the end of six days of waiting, the latter was finally put online, without explanation.
elections at risk
Brazil is one of the countries identified as “at risk” for Facebook disinformation campaigns and in theory of additional specific means to detect and counter destabilization operations. Solicited by Global Witness, the company claims to be “deeply engaged in the protection of elections in Brazil and in the rest of the world”, but gave no details on the reasons why these obviously prohibited advertisements could have been published.
In previous testing campaigns, Global Witness had shown that hateful messages in Burma, Ethiopia or even calls for violence in Kenya were easily passing between Facebook moderation meshes.