“He was a good boy, he will be missed", said a tweet about the teenager who stabbed a teacher to death at a school in São Paulo on March 27, 2023.
If it were the only tweet, it would already be something bizarre, but the publication is far from being an exception: an intensive social media monitoring by Nucleo found hundreds of posts in the last week on Twitter that explicitly extol school massacres and glorify their perpetrators.
After Elon Musk took control of the company and fired over a third of its employees, eliminating a large part of its moderation team, this type of extremist content apologizing for crimes against children and school employees has circulated freely on the social network.
The company's Policy on Sensitive Media, last updated in Jan.2023, prohibits the production and circulation of content with real images of violence.
Nucleo was unable to contact Twitter in Brazil, since the company has no press officers here anymore and the general press email returns only a poop emoji.
Nucleo has been monitoring hundreds of profiles from this Twitter community of school violence aficionados since late 2022.
In the last 10 days, in which Brazil witnessed two school massacres resulting in five deaths, four of them small children, only three out the 160 active profiles monitored suffered some form of moderation by Twitter.
With the lack of moderation from the platform, the main form of containment noted by the monitoring was the public dissemination of many violent profiles, either through press reports or by other users on social networks.
This exposure caused many participants to flee to other platforms, such as Discord or Telegram, or even lock or completely delete their accounts.
FEAR OF EXPOSURE
Those profiles that glorify massacres and criminals prefer to stay under the radar on Twitter, posting content only for members of the community.
After being exposed, some members launched a general alert for participants not to interact with newly created profiles, who ask “too many questions” or use hashtags released by the press, for fear of being identified and denounced by authorities.
To identify themselves, most use specific keywords, less obvious hashtags or sets of numbers that have some representation in the community.
The fear of regular users, according to posts seen by Nucleo, is that, on Twitter, it is impossible to know who you are interacting with, unlike platforms like Discord, where remaining anonymous can be more difficult, since users are constantly interacting with each other on voice or video channels.
The author of a massacre in Buffalo, New York, in 2022 used a private Discord server to announce the attack that resulted in 10 victims, in addition to sharing his supremacist manifesto. Other similar cases have already been reported.
At the time, the platform banned the criminal's account and removed the private server, archiving the content for investigation of users who interacted with the server. The material was also shared with US authorities.
The motivation to use Twitter, on the other hand, is that it's easier to find content and start conversations there.
Despite the “migration” to platforms, our reporting noted that publications have become increasingly violent since the case of the school in Vila Sônia, a neighborhood in Sao Paulo, one of the biggest cities in Latin America.
It was unusual to see tweets with explicit content extolling or planning massacres – for example a tweet in which a user says he wants to commit a crime in his school, showing photographs of the facilities.
Thus, more recently, publications of this type have become the standard of the community, which previously used to interact mostly by sharing fancams (videos constructed with the purpose of praising an individual or group of people).
The most recent posts have also raised the tone of hate speech against women.
In one of the most violent tweets identified by Núcleo, a user describes how he would carry out a massacre at his school and later practice necrophilia with the bodies of the fellow students. In the tweet, which is still active, there are at least five comments encouraging the act.
Community members also began an active search for the formation of “pairs”.
Brazilian authorities fighting extremism say, in relation to previous cases, that perpetrators of attacks on schools usually act in pairs with people they know through social networks or online forums.
IN SEARCH OF WEAPONS
Nucleo also identified community users in a search for firearms, explosives and other types of bladed or lethal weapons.
Others began to publish pictures with weapons. These photos are usually accompanied by skull-painted masks, commonly used by many school shooters and community members, which are a symbol of white supremacy originating from American neo-Nazi cells.
In one case, we saw a user — who identified herself as a woman, including posting videos of her face without any cover — who posted an image of a .38 caliber pistol allegedly in her possession. The user also published, on the same day, a photo of a wall with an AK47 pencil drawing.
“It turned out pretty, my AK, it would be better if it were real to kill these motherfuckers, little do they know that in less than 2 months they will have their heads exploded…”
It is not possible to verify or confirm the veracity of the images. Although some contain faces of the alleged users, they may have been generated by artificial intelligence or taken from groups in applications, and therefore, they may not be found through a reverse image search on Google.
In addition to tweets encouraging school shootings, glorifying shooters, and other types of extremist content, we found community accounts sharing images of graphic violence without censorship or edits.
On more than one occasion, Nucleo found members of the Brazilian community sharing videos from the Islamic State group (ISIS) that contained images of beheadings, beatings and other types of explicit violence.
Although this type of content violates Twitter's Community Guidelines, it has not been moderated.
In its Sensitive Media Policy, Twitter defines graphic content as “any form of media that depicts death, violence, medical procedures or serious physical injury in graphic details.”
HOW WE MADE THIS
Núcleo created a profile to interact with community members on Twitter. We analyzed 160 profiles over a period of approximately four months. We then sent an email to Twitter advisory, but has received no response.