Instagram goes back and removes fake news tag in post by Carlos Bolsonaro – Applications and Software – Tecnoblog

After tagging two almost identical posts by Carlos Bolsonaro (Republicanos), Rio councilor and son of President Jair Bolsonaro, as “false information”, Instagram decided to go back and removed the fake news tag from the politician’s publications. Both featured excerpts from the speech of former president Luiz Inácio Lula da Silva about the sale of Petrobras refineries in Bolivia.


Carlos Bolsonaro had posts classified as “false information” by Instagram (Image: Carlos Bolsonaro/ Facebook)

Instagram removes tag on posts by Carlos Bolsonaro

Instagram put the false information tag on Sunday (7) and reversed the decision on Monday afternoon (8). At first, publications were classified under the “false information” mark.

Carlos Bolsonaro’s posts cited a speech by Lula made in a lecture at the former president’s institute in 2015. In them, the councilor linked the rise in cooking gas — which rose 7.2% in October — to the nationalization of Petrobras refineries in Bolivia, during the government of Evo Morales, in 2007.

The “false information” tag was based on a verification of the to the facts, a vehicle specialized in disinformation checks. The article pointed out that Lula had not simply given the refineries to the Bolivian government, but had sold them for US$ 112 million. There is no proven link between the nationalization of Petrobras facilities and the increase in cooking gas.

Tagging was by mistake in Instagram AI

How did the report of the Techblog, there was a mistake in marking Carlos Bolsonaro’s posts. Instagram, like Facebook, uses an artificial intelligence system to help moderate content on networks. The platform considers checks made by vehicles such as Aos Fatos, or Agência Lupa.

In the case of the video published by the president’s son, the AI ​​stored Lula’s video to use it as a reference to catch “variants” of fake news.

It turns out that the original video had the caption: “Believe it or not. Lula confesses that he gave the Petrobras refinery to Bolivia”. The version published by Carlos Bolsonaro did not contain the same statement, so it could not be taxed as fake news for being “political speech” — Instagram does not moderate this type of speech on the social network.


Post by Carlos Bolsonaro about Lula with the “false information” tag (Image: Reproduction)

Carlos Bolsonaro celebrated the suspension of the “false information” tag from his Instagram posts. In his account on the social network, the councilor says that the platform will reinstate all publications “mistakenly” labeled as fake.

Facebook has already withdrawn punishment from another son of the president

“The post just doesn’t violate Instagram rules,” said people familiar with the matter. A similar case has already occurred with another son of the president: Eduardo Bolsonaro.

The deputy was banned from Facebook for “mistake” according to internal sources. He republished images with quotes from the Nazi dictator Adolf Hitler. Later, on the same day he was suspended, the federal deputy had his profile restored by the social network.

On her Twitter profile, the founder of Agência Lupa, Cristina Tardáguila, mentioned that the AI ​​used to moderate content from politicians — which checkers are not authorized to verify — is “very lame, especially in Portuguese”.

Tardáguila stated that improving AI is fundamental for Instagram:

“Improving artificial intelligence is an indisputable step. Ending the protection of politicians, another. What you can’t do is stay in the middle of the road because this gives rise to a tsunami of hatred against the #factcheckers. Who decides whether or not to mark the politicians is in the US.”

See what Instagram said

O Techblog sought Instagram to comment on the suspension of appointments in Carlos Bolsonaro’s post. The company told the report:

Each post can have thousands of copies on Facebook and Instagram, which may have slight differences, like a different photo clip.

We use machine learning technology to detect these copies so that fact checkers can focus on new information.

In the specific case, the video posted was a different version of the one marked as fake by Instagram and Facebook’s partner fact-checking agencies in July, and therefore the marking was removed.

Leave a Comment