Abstract
The popular encrypted messaging and chat app WhatsApp played a key role in the election of Brazilian President Jair Bolsonaro in 2018. The present study builds on this knowledge and showcases how the app continued to be used in a governmental operation spreading false and misleading information popularly known in Brazil as the Office of Hatred (OOH). By harnessing in-depth expert interviews with documentarians of the office’s daily operations—researchers, journalists, and fact-checkers (N = 10)—this study draws up a chronology of the OOH. Via this methodological approach, we trace and chronologize events, actions, and actors associated with the OOH. Specifically, findings (a) document the rise of antipetismo and disinformation campaigns associated with attacks on the Brazilian Worker’s party from 2012 until the election of Bolsonaro in 2018, (b) describe the emergence of the OOH at the heels of the election and subsequent radicalization in WhatsApp groups, (c) provide an overview of the types of disinformation that are spread on the app by the OOH, and (d) illustrate how the OOH operates by mapping key actors and places, communicative strategies, and audiences. These findings are discussed in light of ramifications that government-sponsored forms of disinformation might have in other antidemocratic polities marked by strongman populist leadership.
The encrypted chat app WhatsApp has been widely used to spread political disinformation in Brazil, as well as in other countries like India (Banaji et al., 2020), Indonesia (Baulch et al., 2022), and Chile (Valenzuela et al., 2021). The 2018 Brazilian elections became a notable example of the political power of WhatsApp, with evidence pointing to disinformation campaigns that favored the victory of President Jair Bolsonaro (Evangelista & Bruno, 2019). The term disinformation is here specified as the clear intention to spread false information to manipulate public opinion (Lazer et al., 2018). After the 2018 elections, WhatsApp continued to be used to spread disinformation in Brazil, harming democratic discourse. However, there was one significant difference: Those who coordinated disinformation campaigns were now located in a room on the third floor of the President’s workplace, in what is popularly known as the Office of Hatred (OOH) (Arbex & Uribe, 2019).
The OOH engaged in government-funded computational propaganda, employing a combination of people and bots to shape discourse—not only on WhatsApp but also on Twitter, Facebook, YouTube, and blogs, to artificially enhance support for the government and to target political opponents. This aligns with emergent understandings of false information being spread in cross-platform campaigns (Lukito, 2020), necessitating explorations of WhatsApp’s role in larger media ecologies (Davis & Straubhaar, 2020; Zuckerman, 2021). The OOH pushed narratives in favor of Jair Bolsonaro’s denial of the COVID-19 crisis; spread disinformation about risks, symptoms, cures, and prevention; and was also responsible for attacks of public figures that supported social isolation (Ricard & Medeiros, 2020). Furthermore, evidence suggests that the OOH systematically spread disinformation to discredit news outlets and journalists that reported about sensitive topics within the government, such as corruption schemes, the deforestation of the Amazon Forest, and the connections between the Bolsonaro family and militias (Nemer, 2020). Investigating the OOH is important for developing a more nuanced understanding of how computational propaganda operates in Brazil. Taking cues from Brazil may also provide insights for other Latin American countries (Pérez & de León, 2021), as Bolsonaro is but one among a larger group of current and former authoritarian populist leaders defying democratic principles.
Prior research has found evidence of the usage of WhatsApp in the context of the 2018 election in Brazil (Evangelista & Bruno, 2019). However, far less attention has been paid to how WhatsApp, after the election, became a political weapon in the hands of the Bolsonaro government—a gap this study tries to fill. We investigate how the messaging application was used during the 2018 elections, the antecedents that culminated in widespread usage of WhatsApp, and how WhatsApp became a tool for institutionalized computational propaganda in Brazil. The study draws on expert interviews conducted with prominent researchers, journalists, and fact-checkers knowledgeable of the political disinformation ecology in Brazil, as well as with victims of harassment campaigns that were coordinated, at least in part, on WhatsApp. Our study showcases how the OOH operated a distributed strategy of propagating disinformation.
Literature Review
Computational Propaganda and Populism
Computational propaganda is the collective use of autonomous agents (bots), social media platforms, and big data with the goal of manipulating public opinion (Woolley & Howard, 2016). For example, bots can be armed with big data about people’s behavior and then used in social media to engage the public on political issues and move forward ideological endeavors. At the same time, disinformation campaigns and computational propaganda often emerge out of networked, collaborative settings (Krafft & Donovan, 2020; Marwick & Lewis, 2017; Wanless & Berk, 2020). In many cases, collective organizing and then tapping infrastructures of genuine humans aid the propagators of disinformation. Therefore, computational propaganda can be seen as a blend of humans and bots to artificially manipulate political communication (Woolley & Howard, 2018). This type of computational propaganda has been reported in India (Neyazi, 2020), a scenario not dissimilar to the Brazilian case. The evidence around the OOH includes automated elements, such as the usage of bots on Twitter (Santini et al., 2021) and the automatic creation of sock puppet accounts on WhatsApp (Mello, 2018). At the same time, news articles have documented human troll armies working to spread disinformation on WhatsApp in Brazil (Chagas et al., 2019). Expanding their definition of computational propaganda, Woolley and Howard (2018) also include digital marketers that manage junk news factories that spread disinformation targeting opponents. Such a definition aligns with the work of the OOH and its endeavors during the 2018 Brazilian election, and of conceptually understanding disinformation, in Ong and Cabañes (2019) terms, as a “culture of production” (p. 5773).
Computational propaganda strategies are diverse: for instance, governments have harnessed political bots to push messages and harass opponents, political campaigns have used computational propaganda to swing votes and defame the opposition during elections, and political actors have organized disinformation campaigns to attack human rights supporters and journalists (Woolley & Howard, 2018; Howard, 2020). Encrypted messaging apps often play a key role in these efforts (Banaji et al., 2020; Rossini et al., 2021). While OOH activities may be conceptualized as a coordinated campaign (e.g., Lukito, 2020), the OOH also invites participation from people who want to join in on the hatred that is spread, akin to the notion of “participatory propaganda” (Wanless & Berk, 2020). Such propaganda typically involves the “opportunistic involvement of community members” (Lewandowsky, 2022, p. 328).
Computational propaganda is an activity frequently associated with digital populism. Populist leaders typically proclaim to settle a crisis in the name of the people, create antagonistic camps, foment the division between “us” and “them,” purport to represent the “real” people via moral claims, and wager emotions toward these ends (Bulut & Yörük, 2017; Mudde, 2004). Oftentimes, populists also attack the media (Canovan, 1999). Digital populism, then, is the concept of populism harnessing the idiosyncrasies of digital platforms, and a useful term to describe Jair Bolsonaro’s political style in Brazil (Cesarino, 2021). Gerbaudo (2018) argues that the success of populist ideas on social media is connected to platforms providing anti-establishment voices a stage. Researchers in Norway found that populist right-wing politicians are perceived as more authentic, relatively speaking, pointing to how populism can serve as a productive communication strategy for politicians (Enli & Rosenberg, 2018). Other recent and contemporary examples of populist politicians who employed computational propaganda in Latin American countries are Nicolás Maduro and Hugo Chávez in Venezuela (Pérez & de León, 2021) and Rafael Correa in Ecuador (Bradshaw & Howard, 2017). In the following sections, we detail how the digital populist Jair Bolsonaro has been using computational propaganda to defend his government and attack opponents, especially through WhatsApp.
Important to note, this study does not argue that the OOH is the first computational propaganda enterprise to ever exist in Brazil. Both the center-left and center-right have deployed computational propaganda in their political campaigns at least since 2010 (Arnaudo, 2018; Bradshaw & Howard, 2017). However, 2018 was the first time that a far-right president was elected in the country. It was also the first time that a president used a digital populist tactic such as creating an institutionalized governmental propaganda department devoted to the spread of online disinformation and systematic assassination of reputations (Nemer, 2020). Recent Brazilian history has not seen comparable accounts of state-sponsored computational propaganda efforts perpetrated that led to simultaneous official investigations by the highest instances of power in Brazilian politics: the Supreme Court, the Superior Electoral Court, and the Congress.
General Trends in Political Use of WhatsApp
WhatsApp has been used in favor of populist political projects around the world (e.g., Jakesch et al., 2021). Its technological features have been shown to facilitate the spread of misinformation (Neyazi et al., 2021). On WhatsApp, communications are protected through end-to-end encryption, which prevents fact-checking within the platform (Kazemi et al., 2022). In conjunction with the lack of a news feed, this also prevents information source traceability. Furthermore, message exchanges often occur between close contacts, given that accounts are associated with a cell phone number (Gursky et al., 2022; Rossini et al., 2021). Increasingly used for news sharing, WhatsApp can also be seen as a form of meso news-space where news engagement occurs inside a more intimate forum of communication (Tenenboim & Kligler-Vilenchik, 2020). Another important affordance of WhatsApp is the possibility to send both interpersonal and group messages. Some users might share political content in large public groups which then gets passed along in family groups and interpersonal messages (Gursky et al., 2022). This creates an environment prone to the capillarization of populist ideas, meaning that content gets networked and spreads between groups and interpersonal chats (Cesarino, 2019). By the same token, public groups related to politics often share memes and other political content that originates in other social media platforms (Resende et al., 2019). All these affordances make WhatsApp unique when compared to other, more public social media platforms.
According to the most recent data available, WhatsApp has more than 165 million users in Brazil—more than 75% of the country (Mari, 2022). The popularity of the messaging application is at least in part due to zero-rating policies that offer free use of WhatsApp in pre-paid plans; that is, users, even those with only pre-paid, limited minute accounts, have free access to WhatsApp above and beyond their paid minutes. In Brazil, such plans are offered by telecom operators and are frequently used by people with lower income (Evangelista & Bruno, 2019). Seventy-five percent of the Brazilian population has access to the Internet and 41% get their news through WhatsApp (Newman et al., 2022). Political disinformation spread through WhatsApp therefore has the potential to reach large audiences.
The 2018 Brazilian Elections and the Office of Hatred
In the end of 2018, right-wing extremist Jair Bolsonaro was elected as President of Brazil. As Davis and Straubhaar (2020) point out, the election was strongly characterized by “antipetismo,” referring to a campaign that produced a highly personal resentment against the Workers’ Party (“Partido dos Trabalhadores”), the party that ruled the federal government from 2002 until 2016. This anti-Workers’ Party atmosphere was fueled especially during huge popular protests in 2013 and culminated in the impeachment of President Dilma Rousseff in 2016 (Davis & Straubhaar, 2020; Mourão & Chen, 2020). Jair Bolsonaro’s election in 2018 can be seen as a continuation of this dissatisfaction, stimulated by disinformation campaigns that were spread through WhatsApp.
Content supporting Jair Bolsonaro on WhatsApp was particularly prone to carry false information: The Guardian reported that, of a sample of 11,957 messages spread within 296 WhatsApp groups during the 2018 Brazilian elections, nearly 42% of right-wing content was false, while less than 3% of the left-wing messages contained falsehoods, as verified by fact-checkers (Avelar, 2019). One notorious example of disinformation that circulated during the 2018 Brazilian elections was what Jair Bolsonaro’s campaign referred to as the “gay kit” (Davis & Straubhaar, 2020; Feres & Gagliardi, 2021). Jair Bolsonaro publicized at various points in time a homophobic, distorted perspective of a set of educational materials that was produced by the governmental Education Department in 2004, during the Workers’ Party government, with the intent to equip educators of public schools with materials for trainings against homophobia. The Bolsonaro campaign reframed, in public communication, that the material would promote homosexuality among children via a “gay kit” (Martins, 2020).
A couple of weeks before the runoff election, The New York Times reported that researchers analyzed 100,000 political images that were circulated in 347 WhatsApp groups, finding that only 8% of the 50 most shared images were truthful (Tardáguila et al., 2018). In September 2019, the two largest newspapers in Brazil reported on the existence of the OOH. The newspaper O Estado de São Paulo was the first outlet to widely spread the term, by reporting that Jair Bolsonaro’s former aides gave that name to the group responsible for the president’s digital media (Rosa & Monteiro, 2019). The name Office of Hatred was coined by dissenting members of the government because of the aggressive tactics used against adversaries. According to news reporting, the Office consisted of an “ideological bunker” (Arbex & Uribe, 2019), installed a few steps from the presidential office, in a room on the third floor of the Planalto Palace—the president’s official office in Brasília. Also in 2019, a parliamentary inquiry surfaced testimonies establishing evidence that Jair Bolsonaro, after the 2018 elections, had decided to keep the computational propaganda machine as an institutional arm of his government, paid for by public resources (Ricard & Medeiros, 2020).
Computational Propaganda via WhatsApp
Nemer (2019) argues that in the first year of Jair Bolsonaro’s presidency, some groups of supporters were divided into new coalitions, due to conflicts over different expectations with the new government. Some coalitions became extremely radical, spreading content that he identified as pro-gun, racist, anti-LGBT (lesbian, gay, bisexual, and transgender), antisemitic and attacking the Northeast region of Brazil, which is heavily Afro-descendant. At the same time, other coalitions were disappointed by Bolsonaro and revealed tactics used when they were supporting the election campaign. For example, several members reported receiving money to spread content. The same former insurgents also revealed the existence of what is referred to as the “Virtual Activist Movement” (MAV). The group had been paid to infiltrate WhatsApp groups and spread pro-Bolsonaro disinformation (Nemer, 2019), serving as an example of how computational propaganda makes use of human troll armies.
More recently, Nemer (2020) denounced a harassment campaign against Patrícia Campos Mello, an investigative journalist who had reported about businessmen that illegally hired marketing agencies to attack the Workers’ Party. Her story about the disinformation campaign on WhatsApp caused investigations with the potential to nullify Bolsonaro’s election. Following the publication of this story, Campos Mello became the target of a disinformation campaign that was widely spread in WhatsApp groups. Some of these groups were managed by the phone numbers of two of Jair Bolsonaro’s sons, Flávio and Eduardo (Nemer, 2020). The Brazilian press has also reported that several WhatsApp groups used during the 2018 elections were still active after the election (Militão & Rebello, 2019). But despite journalistic and scholarly inquiries into the role of the OOH in the spread of disinformation campaigns via WhatsApp in Brazil, layers of opacity continue to cloak how exactly the OOH operates. Against this background, we ask the following research question: How do Brazilian experts, fact-checkers, and journalists describe the computational propaganda techniques and activities of the Office of Hatred?
Method
This study presents a chronology of the Office of Hatred by way of expert interviews. Researchers, journalists, and fact-checkers are deeply committed to the investigation of the spread of disinformation through WhatsApp in Brazil. Journalists, in particular, are considered “interactional experts” (Reich, 2012) as they gather information and accumulate expertise by investigating causes and interviewing others. Against this background, and in combination with researchers and fact-checkers, they form a potent population of interviewees uniquely qualified to help us answer our research question. As a matter of course, the range of voices represented in this study (professionals focused on tracking and uncovering the spread of disinformation stemming from Jair Bolsonaro’s administration and supporters), and the absence of pro-Bolsonaro voices, shapes the narrative of this research. It is equally important to note that the Bolsonaro administration has denied the existence of the OOH (Mello, 2020). We rely on expert interviews here since experts have at their disposal an “advantage of knowledge” (Meuser & Nagel, 2009, p. 18). The OOH is strategically ephemeral (Ringel & Davidson, 2022; Welsh, 2020) in that its existence is difficult to prove, and intentionally so. Therefore, harnessing the expertise of those who have attempted and/or succeeded in tracking and tracing its existence—journalists, fact-checkers, and researchers—becomes a key point of access and inquiry.
Interviewees were selected based on an exhaustive search of individuals who have been contributing to the untangling of the disinformation phenomenon in Brazil. We combed through newspaper articles, academic research, and reports to arrive at a first list of contacts. This was combined with snowball sampling (Handcock & Gile, 2011) as some of our interviewees provided references to other useful sources. Some of the researchers and journalists that we recruited were victims of targeted harassment strategies themselves—such as exposure of personal information (doxxing), defamation campaigns, and death threats—received after reporting on the activities of the OOH. The study received approval from the Institutional Review Board (IRB) of the University of Texas at Austin on 31 October 2019.
Interviews (N = 10) were conducted in Portuguese via Zoom by a study author between June and August 2020 and subsequently translated. Participants were Brazilian citizens, three women and seven men, ranging between 33 and 60 years of age. The interviews lasted between 30 and 65 min, with an average of 45 min. The sample consisted of six researchers, two fact-checkers, one investigative journalist, and one victim of harassment propagated by administrators of WhatsApp groups designed to spread pro-Bolsonaro disinformation during the 2018 Brazilian election. In the “Results” section, we applied pseudonyms to obfuscate participants’ identities for their safety. Interviews were conducted following a semi-structured interview protocol. We harnessed coding techniques used in grounded theory to parse through more than 30 pages of single-spaced transcript and memo data (Strauss & Corbin, 1997). A first round of open coding permitted us to associate abstract concepts with individual quotes in the data. In a second step of axial coding, we parsed through our data to identify relationships between codes and overarching dimensions. Through this synthesis, we arrived at four themes that we use to explain the OOH. We structured the narrative alongside those themes to present a chronology of the OOH.
Results
Based on our interviews, we identified four themes to productively discuss the Office of Hatred: “Before the 2018 Election of Bolsonaro,” “From the 2018 elections to the Office of Hatred,” “What type of content is spread through WhatsApp?” as well as “How does the Office of Hatred work?” We begin in 2012, when prominent evangelical politicians adopted the “gay kit” narrative, over the rise of Jair Bolsonaro and his election in 2018, up to and including the establishment of a governmental office specialized in computational propaganda and its activities. In addition to constructing a timeline, interviews help describe the type of content that is being spread by the OOH—largely attacks on opponents and defense of the government. Furthermore, the operation of the department is examined by mapping different actors involved with it, how they are organized and financed. Finally, the article describes the types of audiences who tend to be more susceptible to computational propaganda efforts from the OOH. Table 1 summarizes the chain of events that led to the establishment of the OOH, according to the perception of Brazilian experts, fact-checkers, and journalists interviewed in this study.
Timeline of Events That Led to the Establishment of the Office of Hatred.
Before the 2018 Election of Bolsonaro
The spread of disinformation through WhatsApp did not begin in the 2018 elections. To understand the path from the 2018 elections to the establishment of the OOH, it is necessary to observe the broader historical context that led to the strategies used by this center of computational propaganda institutionalized in the Brazilian government.
One interviewee, the researcher Leonardo, had been using social network analysis and ethnography to understand the environment of Bolsonaro supporters on WhatsApp. He recalled that, in the days before the 2014 elections, WhatsApp was used to spread a false news article attacking the Workers’ Party, stating that the party had poisoned Alberto Youssef while he was in prison (CartaCapital, 2014). Youssef is a political figure that was in prison from 2014 to 2016 and played an important role in revealing corruption schemes that occurred during the Workers’ Party government (Fonseca, 2016). Leonardo also recalled that the first time he observed disinformation concerning the “gay kit,” which played a large role in 2018 election disinformation, was in 2012, when prominent evangelical politicians used such false information to support their elections. Jair Bolsonaro joined this group of politicians in 2013 and adopted the “gay kit” narrative.
From the 2018 Elections to the Office of Hatred
According to Humberto, what changed from the 2018 elections to the established OOH was the nature of the content that was promulgated. During the 2018 elections, groups with different causes were aligned in favor of Bolsonaro. These groups defended a range of causes such as gun rights and the arming of the population, attacks on gender affirmation, attacks on what they referred to as “moralism,” and the promise of an end to corruption. Under the Bolsonaro government, content propagated by the Office primarily related to the defense of activities of governmental agencies, departments, and representatives connected to the Bolsonaro family. In addition, Humberto explained that different groups were spreading disinformation in parallel to the OOH to defend their own interests. These groups, many of which emerged in the 2013 protests, had their own sets of claims. But what they had in common was their support for Jair Bolsonaro during the 2018 elections.
After the 2018 elections, the differences between groups of supporters became clearer. Sérgio Moro was the judge who sentenced Luiz Inácio Lula da Silva, president of Brazil from 2003 to 2010 and a founding member of the Workers’ Party, to prison, which prevented him from running in the 2018 elections. After Jair Bolsonaro’s victory, Sérgio Moro became Secretary of Justice of the new government. He resigned in April 2020, accusing Jair Bolsonaro of intervening in investigations by the Federal Police (Lopes, 2020). Roberto was one of the first researchers to monitor pro-Bolsonaro WhatsApp groups. With some of the groups created as early as 2014 still active in 2020, Roberto told us that he observed how the departure of Sérgio Moro was accompanied by an exodus from the base of Jair Bolsonaro supporters and the spontaneous creation of WhatsApp groups supporting Sérgio Moro. The researcher noted that there was a peak in the creation of new groups that had not happened since the 2018 elections. These new and spontaneous groups appear to be much more conversation-oriented than prior iterations, having legitimate political debates. Roberto explained that the pro-Bolsonaro groups rarely had conversations, consisting of a space of (re)transmission of hyperlinks and memes. The groups of Bolsonaro supporters became more radical after the exodus of Moro supporters.
Now with the OOH established, the political climate in Brazil was already hyperpolarized. Luiz, a researcher of Brazil’s information ecosystem who had been monitoring WhatsApp groups of Bolsonaro supporters since 2018, observed that the content that was posted by them became more radical over time. According to Luiz, although the groups of supporters were smaller, their members were radicals willing to do anything. He explained that they were “an impenetrable group, in which the factual truth no longer matters. What matters is their truth.” Furthermore, he added: “It is a group that we will not win. We must learn how to coexist.” The WhatsApp groups, although less noisy than in the 2018 elections, continued to spread content that harassed opponents and disinformed people.
What Type of Content Is Spread Through WhatsApp?
In describing the nature of what gets shared on WhatsApp, Leonardo pointed out that “you don’t make up a thing about a politician and it is sustained in a vacuum. You take a narrative that is already circulating and make a bridge between the politician and that narrative.” One example Leonardo provided of this strategy was that people had been believing in the “gay kit” since 2012 because the politician and pastor Silas Malafaia had been spreading it since then. The new narrative addition was to link the “gay kit” to Fernando Haddad, a candidate of the Workers’ Party (PT) in 2018. In this sense, the OOH took existing narratives and connected them with antipetismo—hatred for the Worker’s Party. For creating these narratives, Humberto described what the OOH producers pay attention to: “Every disinformation has a bit of truth. It needs to be credible.” Another researcher, Adriana, shared a similar perspective: “Every conspiracy, every [piece of] disinformation has a link to reality. Otherwise, it won’t stick.”
According to researcher Adriana, the government promotes what she described as “virtual beatings,” referring to instigating mob dynamics of harassment of particular individuals. As previously mentioned, journalists in Brazil who report accusations against the government, often—and systematically—become victims of harassment. Researcher Rafael described how he witnessed the phone number of the journalist Patrícia Campos Mello, who broke one of the first big stories about disinformation in the 2018 WhatsApp campaign for Bolsonaro, being spread through WhatsApp, among phone numbers of other prominent journalists. Rafael confirmed that supporters of Bolsonaro have tried to harass reporters who questioned the activities of the presidency. One of our interviewees, Luiz, told us of death threats he had received after reporting on the activities of the OOH.
Journalist Carlos revealed writing a piece about how a small group of people managed hundreds of groups on WhatsApp that were used to spread disinformation. The journalist exchanged messages with one WhatsApp group admin that agreed to provide testimony, but then suddenly disappeared. After a while, the admin broadcasted a video transmission on Facebook in which they showed the conversation with the journalist. The group admin exposed the journalist’s social media accounts and called on the audience to harass the reporter, who was consequently tagged in several posts and received insults.
Another interviewee, Leonardo, revealed that the campaigns of Jair Bolsonaro have directed attacks toward authorities and institutions such as the Superior Electoral Court, universities, and scientists. Sowing distrust in this manner helps promote alternative epistemic authority—like bolstering the previously mentioned group of individuals—and ultimately may lead to more credibility and clout for them. In a similar fashion, interviewee Adriana also described how she had observed large volumes of content shared online attacking mainstream media organizations and universities. These attacks can serve as entry points into exclusive channels and communities where only their own “truth” is accepted. According to Adriana, “it is the structure of online conspiracy theory,” pointing out how the patterns of the OOH were similar to those of the alt-right in the United States.
Interviewee José was one of the first researchers to report the usage of an organized strategy to spread computational propaganda in favor of Bolsonaro. José recalled the disinformation campaign that was launched after the Amazon fires brought smoke to cities distant from the forest, such as São Paulo, in 2019. Pieces of disinformation blamed nongovernmental organizations (NGOs) and even the actor Leonardo DiCaprio for the fires. Fact-checker Teresa recalled that, in the beginning of the COVID-19 crisis, several PDF files of false academic studies were being spread. These files used a scientific content format to spread disinformation concerning the pandemic.
How Does the Office of Hatred work?
To understand the activities of the OOH, it is important to map the different actors involved with it, how they organize, and how they are financed. The office used on the third floor of the Planalto’s Palace does not encompass the entire scope of pro-Bolsonaro computational propaganda being spread through WhatsApp in Brazil. In addition to information coming directly from the government, independent bloggers, influencers, and even regular people spread content supportive of Bolsonaro and in line with OOH content. Interviewee Luiz noted that the strategies used by the Bolsonaro family on social media empower “lone wolves.” In other words, people may feel encouraged to create disinformation and try to make it viral by spreading the self-created content through their personal channels.
Journalist Nina explained how she did not believe that the OOH was limited to the physical office located next to the presidential office. She observed that several aides of pro-Bolsonaro representatives would work with disinformation, along with blogs and right-wing webpages, besides the Bolsonaro family itself. Nina said, “It is an ecosystem. It is possible that the strategy is established by the OOH. They may choose a specific hashtag or narrative. But they need the ecosystem.” She also suggested that the ecosystem includes bloggers and influencers such as Allan dos Santos, Otávio Fakhoury, Evandro Pontes, Leandro Ruschel, and Bernardo Küster, and added: You cannot dissociate the Office of Hatred from all these right-wing influencers, social media pages, and bloggers. And the representatives who still belong to the pro-Bolsonaro nucleus. And the aides who use fake accounts and keep spreading content in their networks.
In other words, the OOH acts within an ecosystem of disinformation spreaders, formed by several different actor groups such as bloggers, influencers, representatives, and aides. As previously mentioned, the OOH uses a broad range of social media platforms to spread computational propaganda. However, WhatsApp has played a central (and outsized) role in disinformation spread in Brazil because of the app’s affordances and the fact that the Brazilian population heavily uses WhatsApp (Davis & Straubhaar, 2020; Evangelista & Bruno, 2019; Nemer, 2019).
Leonardo also outlined that he thought there was a degree of centralization compounded in the OOH, but that other actors had their own structures, especially representatives with a very radical organic support of Jair Bolsonaro. Leonardo added, “The antipetismo is larger than the Office of Hatred. But the systematic way with which this is done, the strategic and frequent use [of computational propaganda] against specific political actors, is a characteristic of the Office of Hatred.”
Nina described how she had been trying to uncover the financing of the network of disinformation producers that support the OOH. She pointed out that there was a myriad of financial resources, including several crowdfunding websites used to launder money donated by businessmen and representatives’ aides. Furthermore, Allan dos Santos, the founder of the website Terça Livre, an outlet that supports Bolsonaro and is being investigated by the Brazilian Supreme Court, is accused of spreading disinformation (Netto et al., 2020). He stated in the Parliamentary Commission of Inquiry that he was financed by a crowdfunding website called Apoia.se. Such websites would be used to launder money because donors’ names can be kept confidential. Nina explained that YouTube channels would gain money from Google Ads and through a YouTube tool that allows for instant donations. Businessmen who support Bolsonaro would ask other businessmen to chip in. Nina pointed out that they would also hire direct marketing agencies, just as was illegally done in the 2018 elections. Finally, representatives would also hire marketing agencies that spread disinformation.
Bulk Messaging and Virtual Militia
In the 2018 elections, the OOH used bulk messaging strategies—the hiring of marketing agencies to spread large quantities of WhatsApp messages. Nina stated that marketing agencies had apparently stopped this practice after being under investigation by the Superior Electoral Court. Leonardo noted that well-targeted bulk messaging was essential for initial impact at a national level. He noted that there were so-called “super publishers,” who are publishing large volumes of content at a very high speed on WhatsApp. Leonardo stressed that there were different strategies for creating WhatsApp groups: Some were created spontaneously and then sold by their administrators. Others were created automatically (which violates WhatsApp terms of service) by segmenting campaign databases. According to Leonardo, the Bolsonaro campaign used such automatic creation during 2018. However, such automated behavior can be detected more easily by platforms when messages are sent out on WhatsApp in certain patterns. Roberto observed that some users systematically published a particular message to the same group four times a day, in different periods of the day. The researcher also observed that, frequently, the same user managed multiple groups—between 8 and 20 different groups.
Issue Publics and Susceptible Audiences
The audiences that ultimately received and spread disinformation through WhatsApp are rather diverse. While audiences generally congregate around shared support for Bolsonaro, different issue publics exist that do not necessarily align with one another. Leonardo explained, An evangelical Bolsonaro supporter does not share the same values as a supporter who wants the police to kill everyone. A military supporter, who wants all schools to be military, is not the same as the liberal supporter who wants to privatize everything.
Incompatible interests would typically not appear together on WhatsApp as individual groups receive information that is directly targeted to them, and which they would ostensibly be more willing and likely to share. Leonardo described how othering and disassociation from the Worker’s Party was key in uniting disparate groups: “Petista [Workers’ Party supporter] is the other.” This means that all these groups see their opponents (the “other”) as “petista.” This elasticity in the concept of “petista” allows evangelicals, militaries, and liberals to rally against a perceived common enemy. Leonardo clarified that this also meant each group would answer differently when asked what “petista” actually meant: “For one group it means pedophilia, ‘gay kit.’ To another group it means economical incompetence. To another it means corruption. To another it means the terrorists who wrote the 1988 constitution.”
Discussion
This study used interviews with experts specialized in the computational propaganda environment of WhatsApp to produce a chronology of the Office of Hatred, the presidential apparatus of Jair Bolsonaro to disseminate disinformation in a distributed and decentralized manner. Previous research investigated the Brazilian 2018 elections that brought Bolsonaro to power, but subsequent developments that further led to the creation of an office distributing state-sponsored computational propaganda have received little attention. This study fills that gap—providing an account of how computational propaganda continued to be used in Brazil after the election. While the focus of this study was the role of WhatsApp, our explorations have transferable ramifications for debates on the use and role of encrypted messaging apps in political communication.
Our interviews with a group of experts paint a delicate picture for those concerned with the quality of the information shaping public opinion in Brazil. Recent history shows the power of narratives in Brazilian politics: The disinformation spread about the so-called “gay kit” survived four cycles of elections: 2012, 2014, 2016, and 2018, and there is evidence that it was still being used as a political narrative by candidates in 2020 (Martins, 2020).
The ecosystem described by our interviewees is the source of pro-Bolsonaro disinformation that spreads on WhatsApp. Actors working with the OOH function as an alternative source of truth, working in opposition to mainstream media (Cesarino, 2021). The origins and sources of messages are not easily traceable on WhatsApp, meaning that recipients tend to associate the information with a friend or family member who shared it. Therefore, the trustworthiness of close social ties on WhatsApp may function as a heuristic to the trustworthiness of disinformation (Duffy et al., 2020; Herrero-Diz et al., 2020).
The presence of memes in pro-Bolsonaro groups, as reported by interviewees in this study, is in line with scholarship pointing out how memes can shape political discourse. Political elites have been using the humorous and visual nature of memes to emulate grassroots content creation. The platform affordances of WhatsApp further facilitate the circulation of memes, which are especially attractive to younger voters of a wide range of socio-economic backgrounds (Baulch et al., 2022). It is worth pointing out that similar types of content shared via WhatsApp could also be found on other platforms such as Twitter and Facebook. However, the means of distribution on chat apps like WhatsApp function differently from public social media—through groups, forwarding, and individual chats.
The content that spreads through the WhatsApp encrypted chat app follows the rules of traditional computational propaganda. It is created with the imperatives of virality and deception in mind. In the Brazilian case, the goal is to attack opponents and defend the government regarding crises such as the devastation of the Amazon Forest or the COVID-19 pandemic. Future research should examine how Bolsonaro carried over many of these disinformation strategies to his communications about COVID-19 and triangulate interviews centering the OOH with analyses of trace data of the office’s digital activities.
Our study highlights how the OOH does not work in a centralized manner, but rather with a networked strategy and disinformation spread by a range of actors. Audiences who receive this content are hyperpolarized and heavily characterized by antipetismo (Davis & Straubhaar, 2020). Subsequently, the resentment against the Workers’ Party then morphs into broader narratives where several political groups label their opponents as petistas. WhatsApp played a fundamental role by allowing pro-Bolsonaro disinformation to reach those different groups. In the context of digital populism, this evidence is not trivial. All those different groups interpreted petista differently, with “petista” essentially serving as a floating signifier for different groups to project their hate upon (Tuters & Hagen, 2020). WhatsApp allowed for the spread of disinformation catering to each of those separate issue publics individually. This constitutes a productive strategy in the playbook of computational, participatory propaganda as displayed in the case of the Office of Hatred. At the same time, the Workers’ Party did not seem to offer effective resistance to disinformation attacks, relying excessively on traditional mass communication for their campaigns and not preparing for political strategy on WhatsApp (Davis & Straubhaar, 2020).
The spread of disinformation through WhatsApp in Brazil is noteworthy and idiosyncratic in that it is rooted in the historical, political, cultural, and social context of the country. It can also teach important lessons that may be applied to other countries susceptible to political disinformation campaigns.
Footnotes
Acknowledgements
The authors thank the interview participants for their willingness to speak with us and for providing critical insights.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study is a project of the Center for Media Engagement (CME) at Moody College of Communication at the University of Texas at Austin and was supported by the Omidyar Network, Open Society Foundations, as well as the John S. and James L. Knight Foundation.
