The global far-right has become known for its intensive use of digital platforms to disseminate its discursive repertoire. In the United States, Trump became known not only for idiosyncratically employing X/Twitter as a personal communication channel that replaces all public communication, but also for allying himself with Big Tech owners, and even creating and maintaining his own social media, TRUTH Social. Furthermore, on several occasions, he has been depicted being rude or offensive to journalists in press conferences and pronouncements (Carlson et al., 2020). In Europe, far-right parties have admittedly used social media as part of their strategy for spreading anti-establishment rhetoric. The situation is no different in the Global South. Countries such as India, Russia, and China have frequently been reported as regions that pose a serious threat to press freedom, according to data from the organization Reporters Without Borders (Reporters Sans Frontières, 2025). In Brazil, since 2018, with the election of Jair Bolsonaro, several studies have sought to better understand the far right’s digital communication landscape, highlighting the particularly successful strategy of mobilizing groups of agitators through private networks, especially WhatsApp (Santos et al., 2019; Piaia and Alves, 2020; Chagas, 2022).

Since 2013, WhatsApp has become a pervasive medium in Brazil, facilitating not only personal interactions but also supporting small businesses and expanding access to information for segments of the population previously marginalized from public debate (Mari, 2024). Brazil is now known for having one of the largest user bases for the platform worldwide (Statista, 2024), and one of the main audiences to utilize specific affordances such as audio messages (Kischinhevsky et al., 2020), which have become an inclusive means for older, illiterate segments of the population, or those with difficulty typing long texts and participating in more textually elaborate conversations. In this context, the country has witnessed the resurgence of an alternative information exchange ecosystem, which has gradually eroded the credibility of mainstream news media and become a key component of Bolsonarism’s discursive strategy (Chagas, 2022).

Media criticism has historically been associated with the Brazilian left, due to the high concentration of media ownership and of its well-known interaction with the military regime during the dictatorial period (Paiva et al., 2015). The second term of Luís Inácio Lula da Silva’s presidency even went so far as to foster and subsidize independent blogs to act as alternative journalists, occasionally refuting or reframing the facts from a frequently positive perspective regarding the government.

More recently, however, the far right has appropriated this criticism and begun to undermine the credibility and legitimacy of the press, perpetrating attacks and placing the media under fire, in addition to betting on an alternative media network that develops in parallel with and parasitizes the mainstream media (Orso and Goulart Massuchin, 2025). In this context, WhatsApp has become a powerful tool for spreading highly biased information originating from these extremist networks (Ozawa et al., 2023).

Mainstream news coverage and even fact-checking initiatives have been largely ineffective in combating this network, as these are well-organized groups who are often media-literate enough to understand that the ideological content they share is controversial; however, they disseminate it anyway because it resonates with their agendas (Mendonça et al., 2023). Another, no less important, factor in this dynamic is that such individuals are immersed in an informational environment with a distinctive news diet. The framing produced by hyper-biased media is often one of the main, if not the only, source of information to which some of these individuals have access, so that their opinions are formed from interactions with this news and with other users who share it. In this environment, questioning mainstream media has become a source of legitimacy and social capital for some of these users, as well as a means of distinguishing themselves from out-groups, since they present themselves as free, independent, and morally superior to those they perceive as constrained by mainstream media.

This study seeks to understand this dynamics behind the delegitimizing of mainstream press and the subsequent emergence of an alternative media ecosystem within far-right private networks. The analysis draws on a database of more than 87,000 messages exchanged in 40 political discussion groups of Bolsonaro supporters on WhatsApp during Brazil’s 2024 elections.

In three distinct stages, the following analyses are conducted: a content analysis combined with descriptive statistics on the circulation and sharing of a sample of 38,000 news links from hyperpartisan and politically biased media shared in these groups; a textual analysis using n-gram co-occurrences to examine references to press members, adjectives, and character attributions; and an interpretative analysis of nominal and individual mentions that target mainstream journalists. Based on these data, the article provides insight into how users perceive mainstream media and how the alternative far-right media ecosystem is structured in Brazil. It contends that a coordinated effort to destroy the reputation of news organizations is carried out through extreme measures, including (a) harassment and political violence against journalists, (b) repeated practices of inciting public hostility toward the press, and (c) the development of a hyperpartisan alternative media ecosystem that systematically disseminates disinformation, hate speech, and political proselytism.

While there are studies that identify a resurgence of violence against journalists in specific contexts within authoritarian populist regimes, there is still a lack of systematic, comparative, and longitudinal research that establishes a clear association between the strengthening of far-right regimes and the higher incidence of attacks on the press and harassment of journalists. This occurs, in part, because the definition and typologies of violence against journalists are not yet settled in the literature (Le Cam et al., 2021), but mainly because, in non-conflictual liberal-democratic contexts, where political polarization and radicalization are among the main challenges to be faced, such violence can manifest in disguised or subtle ways, often difficult to detect, particularly in public spaces (Le Cam et al., 2021).

Even so, since Nerone’s (2008) pioneering effort to conceptualize the phenomenon, an essay that has become a cornerstone for other definitional attempts, violence against journalists has been identified across different regions of the world. It varies according to factors such as the degree of professionalization of news media and journalistic practice, the prevalence of violent attitudes or discourses in society and the political sphere, and the extent to which violent acts reverberate within and shape the public sphere.

According to Nerone, most developed democracies display relatively low levels of violence. His hypothesis for this pattern is that media ownership structures and market conditions discourage ideological confrontation, while professionalization tends to shield journalists from personal attacks. Conversely, since the press is often associated with political or social transformation, journalists frequently become scapegoats for censorship or exclusionary practices, especially in conflictive contexts or those marked by social, ethnic, or political tensions, but not exclusively. Moreover, motivations may include economic or partisan interests, as well as more radical attempts to systematically undermine the press, resulting in long-term influence operations on public opinion, a behavior that, while operating within the formal boundaries of democratic legality, has been widely employed by regimes testing the limits of illiberalism.

In a more recent study, Le Cam et al. (2021) corroborate the view that violence against journalists is a global and multifaceted threat, manifesting in various forms of offenses, including threats, physical assaults, prior censorship, and violent speech, both offline and online. The authors argue that such offenses pose a serious risk to democracy because, beyond endangering individuals’ physical integrity, they produce a collective silencing effect, intimidating the free press, discouraging professionals, and interrupting nascent careers. This dynamic reflects a global crisis of safety and press freedom, generally driven by political or economic interests. Among the main targets of violence are investigative reporters, journalists covering crime or armed conflict, and those dedicated to human rights issues or high-risk environments (Brambila and Hughes, 2019). Yet, violence against journalists is not limited to physical aggression: it also encompasses symbolic, psychological, and institutional forms (Brambila and Hughes, 2019).

Threats, extortion, attacks, persecution, and torture are more explicit forms. However, it is also crucial to consider the denial of access to public information, judicial reprisals through lawfare, retaliation via forced dismissals or budget cuts, precarious or marginalized conditions, revocation of credentials, undue monitoring and surveillance, and cyberbullying as manifestations of psychological or institutional violence that are equally recurrent and effective in producing lasting damage to the professional environment. Additionally, violence against journalists often takes the form of attacks on minoritized groups, notably women, Black people, and LGBTQIA+ communities. The gendered expression of violence against journalists—particularly under far-right regimes—has been one of the most pervasive tactics employed by world leaders to silence both the press and women’s participation in the public sphere (Brambila and Hughes, 2019; Posetti et al., 2020).

Reports from various organizations on violence against women journalists (Posetti et al., 2020; InternetLAB, INCT.DD, Instituto Vero, DFRLab, AzMina, Volt Data Lab, 2022) have highlighted the severe impact of these attacks on the mental health of female journalists, causing stress, burnout, and even absence from work. However, as with other forms of gendered violence (Sobieraj, 2020), the core problem is structural, since the effects often extend beyond direct victims and reverberate through a chain of indirectly affected individuals. In the case of violence agains journalists, this often includes young women reporters who abandon or interrupt their careers to avoid becoming victims themselves. Additionally, some studies emphasize the significant barriers to reporting such aggressions, particularly in the case of discursive violence on digital platforms, where attacks are often conducted anonymously, privately, and in disguised forms (InternetLAB, INCT.DD, Instituto Vero, DFRLab, AzMina, Volt Data Lab, 2022).

When initiated by political leaders, such aggressions can encourage physical violence by non-state actors against journalists. According to Mazzaro (2023), in her study of Chávez’s anti-media rhetoric in Venezuela, the frequency and persistence of these speeches are positively associated with the incidence of physical attacks on journalists and contribute to the long-term normalization of violence, which progressively increases journalists’ vulnerability. Thus, the author argues that even what may initially appear to be a form of symbolic violence can have tangible and damaging effects on democracy and press freedom.

According to Padovani (2022), in the Italian context, journalists face an ethical dilemma when covering far-right political groups and actors, since the visibility given to their discourses and ideas can normalize extremist ideologies. Her conclusion is that, in such cases, journalism must abandon strict neutrality and uphold the defense of democracy through critical coverage. However, in practice, this stance is not always feasible given the routines of news production and the commercial imperatives of mainstream media, which allow political leaders to exploit the situation within the broader logic of political mediatization (Strömbäck and Esser, 2014). The very duties of the profession render journalists vulnerable and expose them to violence.

Another fundamental aspect is that, although disinformation and hate speech constitute an important backdrop for the emergence of hostility toward journalists within far-right digital ecosystems, the literature does not establish a clear or consensual relationship among these phenomena. First, violence against journalists is not necessarily driven by or contaminated with disinformation (Peña-Fernández et al., 2025). While the two may be related, many attacks do not meet the criteria required to be classified as disinformation. Moreover, although violence against journalists and hate speech may co-occur—particularly in cases involving attacks on women journalists—hate speech is more accurately understood as a specific subset of discursive practices that mobilize identity-based reasoning and undermine human dignity (Waldron, 2012). For this reason, not every attack against a journalist constitutes hate speech, at least according to the most widely accepted definitions of this form of discourse (Dourado et al., 2024). That said, even if these distinctions are essential for conceptual clarity, in practical terms disinformation, hate speech, and violence against journalists are often deeply intertwined. As Blanco-Castilla et al. (2022) argue, digital platforms constitute environments particularly conducive to the dissemination of hate speech against media actors, while Relly (2021) shows that harassment and violence against journalists can be understood as outcomes of media ecosystems shaped by disinformation and digital populism.

In Brazil, former President Jair Bolsonaro used to hold informal press briefings near the Alvorada Palace, the official presidential residence. The area was usually open to the public and attracted groups of supporters who not only cheered but also harassed journalists. On several occasions, Bolsonaro verbally attacked reporters or insulted them directly. In one widely publicized case, in February 2020, he made sexual innuendos against a female journalist investigating a scheme involving the mass dissemination of WhatsApp messages during the 2018 elections. A year later, the former president was ordered by the court to compensate the journalist (LatAm Journalism Review, 2021).

Nicoletti and Flores (2022) also emphasize the role of digital media in amplifying political discourse hostile to the press. Ozawa et al. (2024), however, offer a counterpoint, noting that the increase in threats and attacks against journalists during the Bolsonaro administration did not produce a chilling effect or self-censorship, but instead stimulated resilient behaviors and forms of collective resistance among journalists, a phenomenon they describe as “catalyzing effects.”

Even so, violence against journalists represents only one dimension of a broader operation carried out by extremist and authoritarian leaders and groups. Another common strategy is the creation of an alternative, politically hyper-partisan media ecosystem, driven by a network of influencers and private communication channels, what Lewis (2018) calls an alternative network of influence. This type of system relies on anti-media discourse and competes directly with mainstream journalism, presenting itself as an allegedly independent source of information untainted by political or elite interests, while in reality parasitizing mainstream content and offering ideologically compromised framings.

Shortly after being elected president in 2018, Jair Bolsonaro posted messages on his official Instagram, X/Twitter, and YouTube accounts recommending several far-right influencer channels. He repeated this gesture on other occasions, consistently promoting profiles and channels that positioned themselves as alternatives to mainstream media and were explicitly ideological. This practice is equally common among Trump and other populist leaders.

The literature has established the term hyperpartisan media (Rae, 2020) to designate this type of endeavour. We adopt the definition proposed by Chagas and Möser (2025) instead, which describes these outlets more accurately as politically hyper-biased. They reject the term hyperpartisan because such media are not necessarily linked to political parties, and even if partisan were taken to mean “committed” or “engaged,” this would fail to capture their extreme ideological bias. In other words, this is not merely “committed” media, but media that severely distort framing, often to the point of bordering on disinformation, and, in many cases, actively promoting it.

Another key aspect of this definition is that these outlets are often labeled as “alternative media,” forming a kind of parallel ecosystem that positions itself between audiences and mainstream journalism through an anti-media discourse that delegitimizes professional reporting. However, as Ekman and Widholm (2022) notes, these so-called alternative outlets are in fact deeply dependent on mainstream media, since they rely on its production to comment on, react to, or reframe events, often lacking their own investigative or editorial capacity. They are typically semi-professional or amateur, orbiting the mainstream not as an alternative but as a parasitic layer of media.

Some authors nonetheless prefer the term alternative media to emphasize how these outlets present themselves as unconventional sources of information (Holt, 2019). Indeed, while this phenomenon has become particularly salient on the right, it is also observable among left-leaning initiatives.

In Brazil, during Lula’s second administration (2007–2010), the government strategically fostered progressive bloggers as alternative sources of news. At that time, official advertising in print and broadcast media was reduced to channel greater resources toward independent journalists and initiatives, pejoratively labeled “dirty bloggers” by their detractors (Carvalho, 2017). Yet nothing compares to the network built by Bolsonaro beginning in 2018.

Following the pattern observed by Benkler et al. (2018) in the United States during and after the 2016 elections, the hyper-biased right-wing media ecosystem (led by Fox News, Breitbart, and others) operates as a closed system that strongly rejects professional journalism. According to the authors, this results in an asymmetrical media structure, since disinformation and online propaganda thrive far more intensely in the right-wing ecosystem, which is self-reinforcing and resistant to external correction.

Similarly, in post-Bolsonaro Brazil, an informational crisis has emerged from the systematic questioning carried out by hyper-biased media outlets. These outlets not only undermine trust in mainstream journalism but also disseminate political proselytism in defense of conservative agendas, and are often funded by lobbying groups while also monetizing their own channels.

The Brazilian media system has long been characterized by high ownership concentration and strong political parallelism (Paiva et al., 2015). Moreover, although many outlets later regretted it due to subsequent censorship and repression, a number of them collaborated with the 1964 military coup, which installed a dictatorship that lasted over twenty years (Paiva et al., 2015). As a result, the Brazilian press has often been viewed as aligned with major elite business interests and supportive of right-wing agendas, including alleged indirect political interference during the period of re-democratization. In 2013, when mass protests swept the country—echoing movements such as Occupy Wall Street, Los Indignados, and the Arab Spring—protesters chanted anti-media slogans. One of the most emblematic, #globolixo (“Globo trash”), originally directed against the country’s main broadcaster, Rede Globo, was later appropriated by Bolsonaro’s right-wing supporters, in a paradoxical reversal that can only be understood through Bolsonaro’s populist and anti-establishment rhetoric (Orso and Goulart Massuchin, 2025).

More recently, a proliferation of far-right initiatives has emerged, including channels led by militaristic influencers, monarchists, religious figures (especially neo-Pentecostals), conspiracy theorists, and anti-science advocates, forming a diverse, multidimensional front that supported Bolsonaro’s candidacy (Chagas, 2022). Unlike Lula’s earlier network of bloggers, this ecosystem is primarily composed of social media pages and profiles, and news websites. It is also a cross-platform structure, aptly described as a hybrid media system, as defined by Chadwick (2013). YouTube channels amplify content from blogs and websites, which in turn circulate across X/Twitter and Instagram; meme pages thrive on Facebook; and groups and broadcast lists operate on various alt-tech and private messaging platforms. The latter, in particular, play a central role, functioning as hubs that interconnect other services and platforms. Through group discussions and broadcast lists, most news links circulate and gain traction.

Thus, studying and understanding the circulation of these links provides insight into the informational diet of these audiences (Santos, 2021; Santos et al., 2022). However, it is impossible to grasp their significance without recognizing the pervasiveness of private messaging in Brazil’s communicative landscape, especially WhatsApp, which has become one of the population’s primary channels of communication, surpassing the telephone, SMS, and even traditional social media platforms.

WhatsApp was created in 2009 by two former Yahoo! employees, initially intended as a personal status application, a kind of precursor and competitor to Snapchat, which would only be launched in 2011. However, its rapid adoption soon transformed it into a potential replacement for instant messaging services, competing directly with MSN Messenger, which was discontinued in 2014, and Facebook Messenger, which became a standalone app in 2011. In developing countries, where telephone network coverage was still expanding and SMS or long-distance calls remained costly, WhatsApp quickly replaced these services, becoming a sensation in an exceptionally short time and forcing telecommunications companies to include it in their plans and packages (Santos et al., 2019).

In 2014, Facebook acquired WhatsApp for approximately $19 billion. Two years later, the app adopted end-to-end encryption, a privacy technology that ensures only the sender’s and recipient’s devices can access messages in readable form. Allegedly, the company does not store user messages on its servers, guaranteeing secrecy and untraceability for private communication, although there are indications that message hashes remain accessible after transmission (Santos et al., 2019). In Brazil and other countries, due to its popularity, mobile companies introduced zero-rating policies, exempting WhatsApp use from data charges even after users exceeded their plan limits (Chagas, 2022). This practice further accelerated the app’s mass adoption. Today, recent data indicate that more than 98% of Brazilian smartphones have WhatsApp installed, with an estimated 170 million active users (Panorama, 2025).

Chagas (2022) notes, based on national survey data, that between 2013 and 2014 private messaging tools surpassed online social networks among Brazilians, becoming the main means of communication and sociability. According to the Reuters Institute (2018), more than one-third of Brazilians report using WhatsApp to consume news, ranking it behind only YouTube and Instagram, although the data do not distinguish whether YouTube videos or Instagram posts were accessed via WhatsApp links or not.

In everyday life, WhatsApp has become an important work tool, used by small entrepreneurs, journalists, and service providers for direct communication with clients and communities (Mari, 2024). The platform’s infrastructure for automating responses, sending catalogs, and enabling payments has attracted freelancers and facilitated work among professionals operating in informal sectors. The addition of voice-over-IP and audio/video call functionalities not only created direct competition with applications like Skype—once a market leader and later discontinued—but also enabled large segments of the population who are illiterate or face digital literacy barriers to communicate through non-textual means. The circulation of voice messages on WhatsApp represents a form of quick and accessible communication that has also been politically instrumentalized (Kischinhevsky et al., 2020).

From 2017 onward, however, concerns began to mount about WhatsApp’s role in spreading rumors and hate speech. In India, several cases of lynchings and assaults mobilized through private messages sparked public outrage (Arun, 2019). In Brazil (2018) and India (2019), electoral campaigns were marked by the mass dissemination of misinformation. In Brazil, WhatsApp reportedly played a decisive role in Jair Bolsonaro’s presidential campaign, serving as his primary channel for mobilization and message dissemination, while also hosting numerous political discussion groups among his supporters (Piaia and Alves, 2020).

Although WhatsApp’s Brazilian office reports that over 98% of messages are exchanged between two individuals, the remaining 2% have been sufficient to cause significant damage to the democratic system. Since the 2018 elections, WhatsApp has functioned for many users as a parallel system of political communication, coordinated by networks of supporters and professional politicians (Chagas, 2022).

The model was replicated and refined in subsequent elections, including the 2020 local elections, the 2022 presidential cycle, the 2024 municipal contests, and even amid the pandemic. This process consolidated WhatsApp as both a tool of propaganda and electoral engagement and a vector of large-scale disinformation (Evangelista and Bruno, 2019).

Studies show that these networks mobilized different strands of extremist groups through astroturfing campaigns (Chagas, 2022) and were especially responsible for spreading large volumes of anti-science content during the pandemic and beyond, advancing agendas that delegitimized academic knowledge and fostered distrust in medicine (Massuchin et al., 2021; Sacramento and Paiva, 2020). They also propagated anti-environmental and denialist narratives (Santini et al., 2025), as well as anti-media discourses (Horbyk et al., 2021). In the latter case, groups were routinely used to mobilize supporters against journalists and to disseminate alternative, hyper-biased framings. Another important aspect is that WhatsApp is primarily used by these groups not only for conversation, but also for distributing videos, memes, and news links. This underscores its role not merely as a content production platform or social hub, but as a cross-platform hub that articulates the flow of information circulating across various media environments, like what some researchers describe as the YouTube-to-WhatsApp pipeline (Santos et al., 2022). In certain cases, content is produced and disseminated immediately through private messaging in a hyper-biased, just-in-time (Santos et al., 2022) news production chain, designed to deliver information instantly to a receptive user base.

News of protests in distant places like Nepal can be disseminated within minutes. Similarly, reports from small Brazilian towns that receive little national media attention circulate rapidly, making the platform a hyperlocal, peer-to-peer communication system accessible to virtually anyone. Even other private messaging platforms, such as Telegram, lack WhatsApp’s reach and user base, making it a virtually ubiquitous service unmatched in what it offers Brazilian users. Combined with its frequent use to steer discourse, reframe events, and mobilize supporters, WhatsApp has become a platform capable of consolidating a cynical public imaginary toward the press and, ultimately, facilitating symbolic and physical violence against journalists.

Given this, the present study aims to investigate how WhatsApp functioned, during the 2024 municipal elections in Brazil, as a catalyst for attacks on the media within political discussion groups supporting former president Jair Bolsonaro. To this end, the study is guided by the following research questions:

Based on the data collected and the research questions outlined above, the study also seeks to examine the following hypotheses:

Below, we present information about the methods employed in this study and the resulting analyses.

This study adopts a multi-method approach, with an emphasis on an exploratory analysis of data from a database extracted from political discussion groups of Bolsonaro supporters on WhatsApp during the 2024 Brazilian elections. The groups were selected based on two essential criteria: they needed to declare support for Bolsonaro or the Brazilian far-right in the group title or description, and they needed to have at least 50 users. Entry into the groups began with two initial seed groups, and then the sample was scaled up to a larger set of groups using a snowball sampling technique, in which new groups were incorporated into the sample as invitation links circulated in the existing groups. Once the sample stabilized at approximately 250 groups, reaching a point of saturation in the circulation of new invitations, a random selection process was conducted to choose the final sample of 40 groups systematically analyzed in this article. The complete database contains over 87,000 messages exchanged between August and October 2024, a period defined by the Superior Electoral Court (TSE) as the 2024 electoral calendar, when the first and second rounds of the municipal elections in Brazil took place.

The research follows a covert research protocol approved by the research ethics committee under review no. (omitted for peer review). According to this protocol, for safety reasons concerning hostile research environments, there is no previous identification or presentation of the researchers to the subjects. Also, all user data, personal details, and group names are anonymized and omitted during the data analysis. No private user or group data was made public, and all results refer to aggregated data.

Two distinct analytical procedures were performed using this complete database. In the first, all links disseminated through group messages were extracted. A total of 38,095 URLs were found. From these URLs, all corresponding domains or outlets were identified. Thus, links such as youtube.com, m.youtube.com, or youtu.be were uniformly classified as originating from YouTube. Subsequently, each domain in the resulting list was examined to classify it as originating from mainstream media, social media and messaging platforms, online services and tools, hyper-biased media, or others. This classification followed the same parameters previously indicated by Santos et al. (2022).

In the second procedure, the complete message database (N = 87,377) was used to identify nominal mentions of journalists and potential attacks. A list of journalists and mainstream media companies was used to locate these mentions. This list followed parameters similar to those previously used by InternetLAB, INCT.DD, Instituto Vero, DFRLab, AzMina, Volt Data Lab (2022), expanded with additional names and variants, since some journalists are frequently referred to by derogatory nicknames in these groups. In total, 982 names were included, encompassing 20 company names or generic categories (e.g., journalist, media, press). Considering only unique and normalized journalist names—without orthographic or nickname variations—there were 240 distinct journalists. However, not all were actually mentioned in the analyzed messages. It is also important to point out that research based on a closed list implies a number of limitations, since the list is not exhaustive, and not all journalists are outlined as well as not all variants of their names could be found, which may constitute a significant obstacle to detecting these mentions.

From the complete database, 2,682 mentions of journalists and media companies were identified. Although the absolute number of mentions may appear small at first glance, it is important to note that a substantial share of the 87,377 messages analyzed contained only emojis, images, and videos. Consequently, this proportion must exclusively consider textual messages. Moreover, as the literature on political violence and hate speech has emphasized (Sabbatini et al., 2023), even when the proportion of hateful messages is relatively low, their impact is often far more extensive than the quantitative figures alone would suggest.

Next, distinguishing only the nominal mentions of journalists (N = 141), an n-gram co-occurrence analysis was conducted, using as a criterion a window size that included the 20 words occurring before and after each searched expression and retaining only the 10 most frequent terms among these neighboring words (top_k = 10). The results were visualized both through topic modeling and co-occurence network analysis.

This same database of mentions was then used for a qualitative content analysis, classifying messages that contained any form of attack or derogatory reference to journalists, or that sought to delegitimize the work of the press more broadly. The goal was to distinguish between these two forms of violence against journalists, those that are more personal and those that are more institutional.

Finally, following the statistical analyses, an interpretive procedure was applied to the messages containing personal attacks on journalists, aiming to understand the broader context, motivations, and framings activated by these discourses. The results of the three stages described above (descriptive statistical analysis, lexical analysis, and interpretive analysis) are presented in detail below.

From the complete database of 87,377 messages sent between August 16 and October 29, 2024, 70% (N = 61,134) were sent during the first round of the elections, while the remaining 30% (N = 26,243) were sent during the second round. This discrepancy is explained by the fact that not only is the period corresponding to the second round considerably shorter, but also that, of the 5,569 municipalities in Brazil, only 103 have more than 200,000 voters and are therefore eligible to hold a runoff election—of which only 51 actually did so in 2024. Furthermore, only municipal executive positions (i.e., mayors) can go to a second round, since elections for the legislative branch (i.e., city council members) are proportional. Consequently, the number of contested offices in the second round is much smaller than in the first.

Among the days with the highest message volumes in the groups, the following stand out: the day of the first round (October 6), with 2,845 messages (3.26%); the day when Brazilian electoral law stipulated that no voter could be arrested or detained, except in flagrante delicto or by sentence for a non-bailable crime (October 1), a logistical restriction common to all Brazilian elections, with 2,462 messages (2.82%); and the day after the first round (October 7), which marked the beginning of the official second-round campaign, with 2,377 messages (2.72%). The most common period for sending messages was the evening, between 6 p.m. and midnight, accounting for 34% (N = 29,750) of all messages.

Regarding the groups and users in the sample, message distribution was also uneven. The most active group alone accounted for more than 10% of all messages (N = 9,367). Similarly, the user who sent the most messages did so 2,563 times (2.93% of all messages), suggesting a power-law pattern compared to those who participated less frequently. Within the corpus, there is also a small subset of viral messages alongside a much larger set of conversational and therefore unique exchanges.

Users resident in Brazil were responsible for the vast majority of messages (98.1% of the total). Among them, the states contributing most to the flow of messages in pro-Bolsonaro WhatsApp groups during the analyzed period were, in order: Rio de Janeiro (20.4%), São Paulo (15.5%), Paraíba (7.55%), Bahia (6.93%), and Minas Gerais (6.67%).

Across the 87,377 messages, 38,095 links were identified. These links were categorized according to their domains, taking into account the media profiles of their outlets. This classification resulted in the data presented in Table 1. As expected, links originating from social media and private messaging platforms were predominant (63.8%). These were generally links from YouTube, Facebook, X/Twitter, Instagram, TikTok, Kwai, and WhatsApp itself. YouTube alone accounted for more than 21% of all shared links. However, it should be noted that much of this audiovisual content, like that found on other social media, points to far-right influencers and movements, and is therefore highly biased.

While classifying social media links into discrete categories of mainstream or hyper-partisan media may raise methodological concerns, this approach allows for the identification of distinct patterns of information consumption, insofar as following a mainstream media outlet on Instagram differs substantially from accessing news through its official website. Moreover, although a systematic analysis falls outside the scope of this article, a qualitative exploratory examination of the link set, particularly those originating from YouTube, suggests a pattern consistent with findings reported by Santos et al. (2022). In their study, the authors measured the proportion of links to mainstream media channels on YouTube shared in pro-Bolsonaro WhatsApp groups and found that more than 51% of the YouTube links corresponded to hyper-partisan or alternative media outlets, or to individual independent vlogger profiles, whereas only 2.6% directed users to mainstream media channels. They concluded that participants in far-right discussion groups deliberately avoided sharing mainstream news content, even when mediated through YouTube, thereby reinforcing an informational diet largely confined to hyper-biased news sources.

In the sample analyzed in this study, considering all 38,095 links collected, politically hyper-biased media ranked second in Table 1, with nearly 10,000 links, demonstrating its importance within this ecosystem.

Another noteworthy finding is that, right after YouTube, the second outlet with the highest number of shared links is a hyper-biased website called Pensando Direita (Thinking Right), with a 13.9% share (N = 5,280). While Pensando Direita is a relatively modest initiative compared to other large-scale hyper-partisan websites in Brazil and has a limited followers base on social media platforms such as Instagram, it operates through a highly articulated and geographically widespread network of agents within private messaging groups, which results in a strong presence on WhatsApp. Accordingly, although it is not possible to rule out the possibility that, during the period analyzed, the channel owner or specific interest groups temporarily dominated the agenda and circulated Pensando Direita links more intensively than usual, the prominence of this outlet within the dataset cannot be dismissed. Moreover, research on WhatsApp is largely based on non-probabilistic samples which, given the opacity of the platform, preclude precise assessments of what would constitute a representative proportion within the analyzed sample. Thus, Table 2 below lists the main hyper-biased websites according to the number of links shared in the sample, along with their respective brief descriptions.

Politically hyper-biased websites are independent and radicalized media ventures. Some gain significant traction on private messaging platforms through deliberate promotional efforts. But these are heterogeneous outlets: some have a local presence, while others operate nationally; some focus exclusively on politics, whereas others address a broader range of topics. Certain cases stand out for attempting to emulate established outlets—for instance, Terra Brasil Notícias, a right-wing portal whose name alludes to the well-known mainstream news site Terra.

Outlets such as O Antagonista, Revista Oeste, and Pleno News exhibit a more professional structure, employ journalists, and produce original stories. Others, like Opinião ES and Nova Iguaçu 24 h, function primarily as local amplifiers of biased news content. The profusion of hyper-biased links indicates that a genuine ecosystem of alternative right-wing news media is in operation.

This pattern becomes even clearer when comparing the circulation of links from mainstream media within Bolsonaro supporter groups on WhatsApp. Nationally prominent outlets, such as the newspapers O Globo (N = 66), Folha de S. Paulo (N = 13), and Estadão (N = 10); portals like G1 (N = 53), Metropoles (N = 45), and Terra (N = 6); and broadcasters such as CNN Brasil (N = 41), show far more modest figures by comparison. The conclusion that emerges is that Bolsonaro supporters actively avoid sharing links from mainstream media and, when they do, it is typically to criticize the press’s stance or to reinforce a particular framing, paradoxically invoking the authority of reputable outlets to legitimize their own narratives.

For the following analysis, we worked with a smaller sample of 2,682 messages sent to far-right political discussion groups that directly mentioned journalists, news companies, or generic terms related to the press. This total was obtained by searching the entire dataset of 87,377 messages for nominal mentions and variations using a pre-established list.

Regarding these mentions, Table 3 shows that references to media companies are predominant in the sample, accounting for 54.1% of the total (N = 1,451). In second place are generic mentions of terms such as “journalist,” “press,” or “media,” occurring 40.6% of the time (N = 1,090). Finally, nominal mentions of individual journalists represent 5.26% (N = 141). While individual mentions occur on a much smaller scale than the other two categories, it is worth noting that company mentions are equally direct and nominal attacks, as they often refer to outlets perceived as unsympathetic to Bolsonaro (such as Rede Globo), whereas supportive companies, such as Rede Record and SBT, when mentioned, are referenced in a positive manner.

Regarding individual mentions of journalists, the pattern is similar. Some are consistently mentioned in complimentary terms, as allies, while others are invariably referenced pejoratively. Table 4 lists the most frequently mentioned names.

The lexical analysis of terms most frequently co-occurring with these journalists’ names reveals important patterns in the anti-media discourse circulating on WhatsApp. As shown in Figure 1, misogynistic patterns appear in references to journalists such as Fátima Bernardes (one of the most important female journalists in Brazil, who was one of the hosts of the country’s most important television news program for 13 years, often described as Congressman Túlio Gadelha’s girlfriend) and Míriam Leitão (a leading economics commentator, mockingly referred to as leitoa, or “piglet,” in a wordplay). Other recurring patterns involve religiosity (for example, André Trigueiro, an environmental journalist, frequently criticized as a follower of Spiritism); politics (Leonardo Sakamoto, a progressive political commentator, invariably called “communist”); and media-related derision (Malu Gaspar, political journalist and podcast host, associated with globosta—a portmanteau of Globo and bosta, “shit”—and William Bonner, probably the best-known Brazilian journalist and most associated with Rede Globo, having hosted the main Brazilian television news program for 29 years, with globolixo). There are also specific personal cases, such as references to an illness affecting Evaristo Costa, a young former host of Globo, and to the helicopter crash that killed Ricardo Boechat, an experienced and critical journalist with experience in multiple media outlets.

In Figure 2, beyond the co-occurring terms themselves, the semantic proximity between them becomes visible. Here, the centrality of terms such as “journalist,” “reporter,” and “Globo,” along with names like “William Bonner” and “César Tralli,” hosts of the two most important Globo TV news programs, constitutes the main cluster of associations.

Finally, the sample of 141 nominal mentions of journalists was qualitatively analyzed to determine which messages effectively contained attacks. The results show that 34.75% of the messages (N = 49) contained ad hominem attacks, while 22.7% (N = 32) included more general delegitimizations of the press. The conclusion is that systematic mentions of the media, and of journalists in particular, often emphasizing episodes from their personal lives, serve to undermine the credibility of the press, thereby creating space for politically hyper-biased outlets to gain traction.

The data reveal a consistent pattern of antagonism toward the mainstream press in pro-Bolsonaro WhatsApp groups, where journalists and mainstream media outlets become recurring targets of discursive attacks. Moreover, the fact that mentions of media companies and generic institutional terms vastly outnumber mentions of individual journalists suggests that hostility is directed less at specific professionals and more at the institutional figure of the press itself. This pattern aligns with how authoritarian populism frames the media as an elite institution acting against the interests of “the people.”

However, it is worth noting that messages explicitly mentioning journalists not only sustain a logic of personalized attacks but also reproduce moral framings aligned with the far-right social imaginary. Gendered mentions of the economic journalist Míriam Leitão or the derogatory treatment given to Fátima Bernardes fit within patterns of typical misogyny. Conversely, the pity expressed toward Evaristo Costa reinforces a kind of Christian moralism rooted in compassion and justice. In cases such as those of journalists André Trigueiro and Leonardo Sakamoto, there is a symbolic reduction of these figures to simplified caricatures, aligned with other enemies of populist rhetoric, such as spiritualism and communism.

A striking aspect is that the valuation of mentions of journalists depends heavily on immediate, contextual factors. In the cases of Glenn Greenwald and Monica Bergamo, both known critics of Bolsonaro, when their reports or opinions resonate with narratives favorable to the far-right moral, criticism tends to fade. Thus, some journalists are mentioned opportunistically or contingently. For instance, Greenwald, an American journalist and former editor of The Intercept, was frequently attacked by Bolsonaro’s supporters for his investigations exposing the lawfare campaign against then–former president Luiz Inácio Lula da Silva, but is now praised for his combative and independent stance toward Lula and the leftist Workers’ Party.

Anselmo Gois is a well-known columnist who frequently addresses issues related to the city of Rio de Janeiro, portraying not only its urban culture but also episodes of violence that resonate with the right-wing agenda centered on public security. Garcia, on the other hand, is a former Globo anchor who has, in recent years, become a prominent voice in right-wing opinion journalism. In this way, Greenwald, Gois, Garcia, Bergamo, and Cláudio Dantas, are consistently portrayed positively, whereas others are almost exclusively framed negatively.

In all cases, individual journalists function as circumstantial targets, while mentions of journalistic companies and generic media-related terms serve to erode long-term trust between audiences and information providers. This process allows the alternative and politically hyper-biased media ecosystem to emerge as the “authentic” source of truth. The broader picture is one of discursive substitution: the systematic weakening of journalistic authority in favor of ideological loyalty within a digitally mediated public sphere fragmented by misinformation.

To address the previously posed research questions (RQs), this study employed three analytical stages: a statistical analysis, a lexical analysis, and, finally, an interpretive contextualization. Regarding the first question (RQ1), which sought to identify the main news outlets featured in links circulating within far-right political groups on WhatsApp, the data indicate that social media platforms such as YouTube dominate these environments. However, politically hyper-biased websites also play a significant role, not only because they attract more shares than mainstream media but also because they encompass a wider variety of outlets.

The second question (RQ2) aimed to understand the types of anti-media discourses circulating within WhatsApp groups and how they facilitate attacks against journalists. The analysis of co-occurring terms revealed that these attacks fall into three main categories: those directed at individual journalists, those aimed at media companies, and those targeting the press more broadly. Moreover, these attacks reinforce moral frames and rhetorical patterns consistent with the logic of populist anti-media discourse.

As H1 claimed, the statistical data confirm asymmetries in the distribution of shared links. Regarding H2, the interpretive and content analyses suggest that while mentions of the press and journalistic institutions outnumber those of individual journalists, proportionally more attacks target specific journalists than the media in general, thus only partially supporting the hypothesis. In practice, the results point to a combination of personal and institutional forms of violence that collectively foster an alternative media ecosystem characterized by strong ideological bias and the systematic dissemination of disinformation.

This study reinforces the notion that private communication networks such as WhatsApp serve as key infrastructures for the opaque circulation of disinformation and the erosion of public trust in the media. Although it focuses on Brazil and specifically on the municipal elections of 2024, and while the analyses are fundamentally exploratory, the findings shed light on how far-right groups and authoritarian populist leaders instrumentalize anti-media discourse as part of their broader political communication strategy.