CTEC Investigation: White Supremacy, Anti-Semitism, and Violence in Spanish and Portuguese Online Communities
| by Alex Newhouse, Gabriela Zayas-Alom, Sophie Liebel, Paulo Magalhães de Paula, Carles Andreu, and Mike Donnelly
Western media coverage of far-right extremism, especially of White supremacy and neo-Nazism, often focus on English-speaking communities, as many of the most high-profile examples of White supremacist terrorism have taken place in the United States, Canada, and New Zealand. However, as CTEC has shown previously, far-right extremism is truly transnational and trans-cultural.
Over the last few months, CTEC has undertaken in-depth research into Portuguese and Spanish extremism on the Internet as part of our partnership with Spectrum Labs. We have uncovered large, highly engaged, radicalized communities of users who frequently and casually share Nazi propaganda, anti-Semitic conspiracy theories, incitements to violence, and direct calls for extermination of certain minorities. These communities span Latin America, South America, and the Iberian Peninsula, and employ a diverse array of dialects, slang, memes, and slurs.
In addition to posing a challenge for content moderators and the development of detection technologies, the existence of extreme communities on readily accessible websites indicates a dangerous situation that could - and, indeed, has already - spiral into racial violence and terrorism.
Spanish Extremism
While CTEC discovered Spanish-language hate speech across mainstream platforms, unregulated websites not surprisingly contained the most extreme instances of bigotry and violence. Hispachan and 4chan, in particular, provide users anonymity and lack of content moderation, resulting in a high concentration of racism, sexism, homophobia, and anti-Semitism.
Hispachan was created in 2012 for the Spanish-speaking online audience as an alternative to the more English-centric 4chan, and it provides discussion boards on an array of subjects, from country-specific discussions to topics of general interest, such as music and science. However, Hispachan is also filled with hate speech, including racist, homophobic, anti-Semitic, and misogynistic comments. It is common to find this type of speech alongside insults, threats and generally toxic posts. Raw, unfiltered opinions and incitement to violence are also occasionally posted on threads started by users asking for advice on real-world situations in their lives.
As with other regions, Latin America has a long, violent history of racism. Most slurs and hateful comments were directed at Black communities, but anti-Semitic hate speech was also prominent. Image boards, including 4chan and Hispachan, have been saturated with anti-Semitic conspiracism and bigotry for years, and latent animosity towards Jews within some Catholic communities in Latin America likely also contributes to this trend.
CTEC tracked toxic users from across Latin America, South America, and the Iberian Peninsula. The vast majority of users were from Mexico, Spain, Venezuela, Chile, Argentina, and Colombia, and sharing national identities frequently caused targeted harassment and insults based on nationality. Further, much like 4chan in English, Hispachan has developed its own slang and dialect: some of these slang words were adopted directly from 4chan, while others are unique to this platform. Devoid of any other context, much of this slang is benign, such as “kek” (laughing, pulled directly from 4chan). However, others are clearly linked to more aggressive and dangerous speech, including “basado” (based, a slang term adopted from English that is used to refer to people acting against social norms, and is sometimes applied to perpetrators of mass shootings and hate crimes), “jorge” (someone who enjoys violence or believes they are hardcore), “jorgeada” (any kind of violent action), and “onions” (synonym of the verb to be).
Although the density of hate speech and violence is highest on these unregulated sites, it is important to note that such toxic content is not unusual on mainstream platforms like Twitter and YouTube. The reason why threats and hate speech found on regulated websites have not been taken down remains unclear. The nuances and vast linguistic variation in Spanish could account for this issue, as well as the fact that algorithms that find this type of content are much more advanced in English than in Spanish. Hateful and violent speech on regulated platforms also generally employs more sophisticated tactics to evade content moderation, including purposeful spelling errors and the use of symbols and allusions to “mask” the true intent.
Portuguese Extremism
Brazil is one of the most racially and ethnically diverse nations in the world, with mixed-race people accounting for over 40% of the total population. As a result of Brazil’s somewhat unique demographics, Brazilian hateful and violent speech diverges significantly from its American counterparts. CTEC’s research has uncovered dozens of examples of violent bigotry toward Brazilians of African descent, Indigenous peoples, and Jews, as well as many instances of online Brazilians obsessing about bloodlines and ancestry. In the most extreme Brazilian communities, White supremacy and outright neo-Nazism are not unusual. In addition, comments and posts and expressed disregard for other races, hatred, and calls for violence garnered dangerous levels of engagement.
The most overt and explicit Portuguese-language White supremacist community exists on Stormfront, the infamous neo-Nazi Internet forum. Stormfront has a board dedicated to Portuguese and Brazilian users. Brazilian White nationalists and supremacists that gather on Stormfront often refer to each other using “proper” Portuguese, avoiding the use of slang and dialectical markers that might mark them as “lesser” Brazilians. They also frequently have cordial and casual discussions about toxic and dangerous beliefs.
The site remains very active today, with some threads having over 1 millions views. Brazilian Stormfront users pride themselves on their pure European ancestry and some idolize eugenics. In order to legitimize their claims to be part of the White race, many members introduce themselves by quoting their ancestry, and often talk about what they admire most in their life and lineage. To express his White pride, one member introduced himself as a strong supporter of eugenics, racial hygiene, genetics, racial anthropology, and social-nationalism. Another user even referred to Hitler as “the great one,” while a different person remarked on his admiration for the Third Reich. Occasionally, users greeted new members in German and used Nazi phrases.
On Stormfront, users are more concerned with organizing activism and encouraging each other than with recruiting new members, since the website is dedicated to people who have already been radicalized into White supremacist ideologies. Such hate speech often takes the form of users frequently sharing resources that can help educate each other about the importance of the White race and how to overcome challenges surrounding the uncertainty of “being White enough.” They refer and quote White nationalist Richard Spencer, as well as other websites, such as the inactive Brazilian website “Legião Identitaria,” which is an underground movement that seeks to “strengthen the Euro-descendent identity and culture of the Southern Region of Brazil.”
Racial hatred was also expressed in other ways. One thread had to do with the fact that a Black woman won the Miss Universe title. The comments about this Black woman compared her hair to steel wool and remarked at the esthetic superiority of White women. She was called a monkey and one user wrote that if she can win the title then he should have entered his dog into the competition. These users seemed genuinely angry and confused by the fact that a woman who was not White could be considered the pinnacle of beauty.
CTEC has also tracked several other, more specifically Brazilian sites that foster extreme exterminationist communities. “Gray web” sites, such as 55chan and Dogolachan, provide a space for Brazilians to express unfiltered thoughts rooted in dangerous ideologies. Dogolachan is perhaps the most notorious of these, as its creator is the first person in Brazil to be prosecuted and convicted of hate crime on the internet. It is believed that Dogolachan had influence on a domestic terrorist attack in Suzano, a city in the greater São Paulo area, as it was observed that the members of the forum praised the attacker and his actions. While Dogolachan was taken off of the clear web after the attacks in Suzano, the “Dogola” meme and the vitriol of the site inspired copycat boards on other sites like Endchan. Referring to Brazil as “Brownzil” due its diverse culture, people that interact on Endchan’s /dogola/ board often encourage their members to promote acts of violence, sometimes followed by instructions and recommendations on how to affect more people.
Further, the language used in this forum deviates significantly from that of Stormfront. It does not portray any level of formality and uses speech densely saturated the most toxic version of hate speech, as members use slang and memes that are overtly racist, misogynistic, anti-Semitic, and homophobic. This extremist niche also encourages doxing and explicit violence against leftists, Black Brazilians, LGBTQ+ people, and women.
Across the Brazilian gray web, users expressed their disdain and hatred for a variety of different racial, ethnic, and religious groups. Specifically, White Brazilians on these platforms spewed hate-filled rants about Black and mixed-raced people in the country. Many referred to Black individuals as “macacos” (monkeys). One user on 55chan wrote that “killing inferiors (Blacks, Jews, leftists) is always right. It is by eliminating parasites that you build a prosperous nation.”
Posts on Dogolachan and 55chan’s /pol/ board frequently exceed the vitriol and violence of even 4chan’s English /pol/ board, which is itself infamous for promoting race war. While 4chan’s English users usually cast overt racism and violence with at least a superficially satirical tone, CTEC’s analysis found 55chan posters seriously promoting Nazi ideologies and calling for genocide. For example, one post reads: “Genocídio é a solução. Solução política não existe. Extermínio é sempre uma solução viável e aplicável. Enquanto ficarem presos à soluções não violentas, o país afundará.” (“Genocide is the solution. A political solution does not exist. Extermination is always a viable and practical solution. As long as we are stuck with nonviolent solutions, this country will be destroyed.”)
Over the course of our research, CTEC discovered frequent examples of White supremacy and outright exterminationist and accelerationist racism on Brazilian websites. However, the hateful ideas that can be found nowadays on Brazilian websites are not new nor completely foreign. Brazil’s social problems are tied to class structures’ issues and policies that have enabled economic disparity between races and classes, but the underground white nationalist movement is pushing to perpetuate the social differences across races. Brazil has even had experience with eugenics, when after the abolition of slavery White elites tried to encourage to European settlement in Brazil to “Whiten” the country. As a result of this fraught history, contemporary racism and White supremacy come in many forms.
Conclusion
CTEC’s research confirms that White supremacy and neo-Nazism are not unique to English-speaking and predominately White countries. Large exterminationist and accelerationist communities have been developing for years on Portuguese- and Spanish-language platforms. The recent history of racial tensions, hate crimes, and xenophobic populism in some nations throughout South and Latin America has demonstrated that there is a need for a greater international effort to tackle these issues.
However, these communities pose unique challenges for developing content moderation and automated detection capabilities. Achieving full coverage of the rich linguistic diversity of Latin and South America is impossible, but content moderation and AI detector development should aim to include at least some data from across regional and socioeconomic lines. Contextual AI is vital for improving coverage while reducing bias, and companies and policymakers aiming to tackle these issues must be willing to adapt to the rapidly evolving, regionally diverse extremist ecosystems that can inspire real-world violence.