A Cautionary Tale: Brazilian democracy, anti-democratic riots, and Meta’s platforms

Este estudo de caso também está disponível em português

On January 8, 2023, Brazilian democracy was threatened by anti-democratic riots. Following Jair Bolsonaro's defeat in the 2022 presidential election, a mob of nearly 4,000 of his supporters invaded, vandalized, and looted government buildings in the Brazilian capital. For more than three hours, groups of individuals proudly adorned in the colors of the Brazilian flag and an array of military-themed clothing occupied and attacked the Presidential Palace, the National Congress, and the Supreme Federal Court. In a disturbing resemblance to the attacks on the U.S. Capitol, their objective was to overthrow the elected government. Despite its shocking nature, this event came as no surprise, given the abundance of anti-democratic, extremist, and violent content circulating on Meta’s platforms – Facebook, Instagram, and to a lesser extent WhatsApp – in the days leading up to the attacks and during the 2022 elections.

In this article, I explore two connected dynamics pertinent to the Brazilian democracy: Firstly, how Meta, the company behind Facebook, WhatsApp, and Instagram, played a role in the challenges to Brazil's democracy during the 2022 presidential election; secondly, how Meta was involved in the anti-democratic attacks by pro-Bolsonaro supporters on January 8th. Central to these challenges is the pivotal role played by Meta’s companies as a conduit for misinformation and fake news, specifically targeting the Brazilian electoral process and its authorities. This work builds on reports from civil society organizations, news outlets, and data from the Federal University of Minas Gerais (UFMG) WhatsApp Monitor, who have collectively been monitoring Brazil’s internet landscape for years and drawing attention to the threats of private companies – such as Meta – in influencing the countries’ democratic institutions.

The research points toward a complex network of private companies and Bolsonaro’s supporters utilizing Instagram, Facebook, and WhatsApp as channels for spreading misinformation and mistrust about the electoral process. These influence campaigns efforts not only provided ideological support to groups involved in the anti-democratic attacks on January 8th, but also facilitated their organization.

Brazilian Politics and Meta Platforms

To put it shortly, Meta and Brazil ‘go way back’. Facebook, Instagram, and WhatsApp gained widespread popularity in the country during the early 2010s. By 2014, Facebook boasted 62 million Brazilians accessing the platform on a daily basis. By 2023, Facebook's user base had expanded to an estimated 109.1 million. WhatsApp, the sibling platform, also enjoyed immense popularity, with around 165 million users in 2022. Instagram also presented remarkable figures in 2023 and was considered by Forbes as the most consumed social media platform in the country. Considering these substantial figures and acknowledging the diverse range of activities these platforms facilitate, from family communication to business operations, it is reasonable to assert that they have become deeply ingrained in the country’s social fabric. Hence, to better grasp the role of these actors in the 2023 attacks, it is important consider the history of these companies, especially since the 2018 presidential elections, through the 2022 referendum and until the January 8th anti-democratic attacks.

In the U.S. elections in 2016, Facebook played a significant and controversial role by serving as a platform for targeted advertising amidst concerns about the spread of misinformation, the use of data analytics, and potential foreign interference. In the aftermath, Facebook promised to do its best to protect forthcoming electoral processes around the world. In 2018, Facebook became vocal about its interest in protecting the electoral process in Brazil, stating that: “The presidential election in Brazil is a top priority for Facebook and we have been working really, really hard to protect the polls and encourage civic participation.”‐ Meta Blog, 2018

The company partnered with fact-checking agencies in the country and started acting against inauthentic behaviour by monitoring and taking down accounts sharing misinformation or redirecting online trafficking to suspicious websites.

On October 22nd, 2018, just six days before the second round of the Brazilian election, Facebook took action by removing 68 pages and 43 accounts associated with the Brazilian company Raposo Fernandes Associados (RFA), included in the Influence Industry Explorer.

See Raposo Fernandes Associados (RFA) and other companies related to the industry in the Explorer database

This step was prompted by violations of misrepresentation – when a page does not authentically represent who they really are – and spam policies. According to Facebook, the individuals behind RFA had created pages using fake or multiple accounts with identical names, leading to their removal. These pages were utilized to spread click bait, using catchy and misleading titles to direct users to external websites functioning as ‘ad farms’ – websites that create revenue by bombarding users with ads and deceiving them into clicking on those ads. The company’s decision to eliminate these pages was based on the behavior of the actors involved, specifically their use of fake accounts and consistent spam posting, rather than the content itself. Notably, spammers are increasingly leveraging sensational political content across the entire political spectrum to build audiences and drive traffic to their websites, earning revenue for every visitor to the site.

Read more about understanding how and why spammers drive traffic, plus tips on identifying real content from spam in our case study 'Investigating Compromised Websites: The Legacy of Cambridge Analytica'.

The removal of these pages was one of the actions Meta took to limit the spread of misinformation and influence campaigns on Facebook during Brazil’s election. Notably, Instagram did not undergo any specific measures from the company, given its lesser role as a conduit for misinformation. WhatsApp, on the other hand, emerged as one of the primary channels for influence campaigns during the 2018 electoral period. Yet, the company stated that they could not implement any new measures to curb misinformation campaigns as there was not enough time. During the campaign period, around 11,957 viral messages were identified in 296 pro-Bolsonaro group chats on WhatsApp, with 42% of them containing misleading content. Below, we can see examples of the misinformation that circulated during that period, blending anti-LGBTQIIA+ rhetoric with religious undertones (Image 1).

Image 1: The confirmed falsified image above depicts Manuela D’Avila, vice-presidential candidate who ran against Bolsonaro, wearing a shirt that says 'Jesus is trans.' The text below reads 'This is Haddad's vice-president. Take a look at what she wears and have no shame in declaring. Think carefully about who you vote for, so you don't vote for people who mock our God. Remember: you don't play or mock with God.' Read more at Rolling StoneCredits: UFMG WhatsApp Monitor

One of the key tools and services offered by private firms to carry out these political influence campaigns on WhatsApp in 2018 was bulk messaging services. Bulk messaging involves sending a large volume of messages simultaneously to multiple recipients through the WhatsApp platform. According to journalistic sources, companies such as Yacows, Quickmobile, Croc Services, and SMS Market were contracted by third parties to disseminate pro-Bolsonaro political messages in bulk via WhatsApp. The companies used databases belonging to Bolsonaro's party supporters as well as databases with names of voters obtained from other digital strategy agencies to get contact numbers. According to Brazilian electoral laws, companies are prohibited from offering support or services for free to candidates or engaging in the commercialization of third-party personal data. Candidates are only permitted to utilize data that has been voluntarily provided. The media attention led the Brazilian Supreme Federal Court to investigate the alleged violations. In parallel, and potentially in response to the media and the Brazilian authorities, Meta blocked several accounts linked to the companies responsible for bulk messaging on WhatsApp. Despite these measures, the content aimed at influencing the electoral process had already gained traction on social media because the fake content had been circulating across Meta platforms since before the beginning of the electoral process in 2018 and fraudulent content tends to be recycled according to the goal of the distinct political groups.

Another particular case highlights the complex network of companies involved in efforts to influence the outcomes of the 2018 Brazilian elections. A recording featuring Luis Novoa, the owner of two Spanish tech companies, Enviawhatsapps and Wachatbot, came to light in 2019, after the elections. In the recording, Novoa revealed that Brazilian companies had hired a marketing agency in Spain, which utilized his software to send bulk messages to pro-Bolsonaro supporters during the 2018 election. Novoa claimed to be unaware of his software's use in Brazilian political campaigns and only discovered it when WhatsApp suspended the company's phone lines alleging misuse. Novoa had previously been responsible for conducting mass WhatsApp messaging campaigns for the Spanish political parties Podemos and PSOE. WhatsApp suspended accounts associated to the Spanish parties Podemos, PSOE and People’s Party during the 2019 electoral campaign, alleging violations of platform usage rules due to engaging in mass messaging.

In summary, the longstanding relationship between Meta and Brazil, marked by the popularity of Facebook and WhatsApp in social and political life, faced challenges in safeguarding the 2018 electoral process. Despite Meta's actions against inauthentic behavior, exemplified by the removal of accounts linked to Raposo Fernandes Associados, WhatsApp emerged as a significant channel for influence campaigns during the 2018 Brazilian elections. Private companies offering bulk messaging services, such as Yacows, Quickmobile, Croc Services, and SMS Market, played a pivotal role, leading to investigations and measures by the Brazilian Supreme Federal Court and Meta. In the next section, I will explore the role of Meta-owned social media during the 2022 election period and on the January 8th anti-democratic acts.

Brazilian Meta-Politics: 2022 elections and the 8th of January attacks

As expected, following the controversial involvement of Facebook and WhatsApp in the 2018 Brazilian presidential elections, Meta came under scrutiny from civil society actors and the media. In response, a series of partnerships were established between Meta and the Brazilian Supreme Electoral Court for the 2020 municipal elections. This collaboration included the development of a WhatsApp chatbot, providing users with access to verified news and clarification about the electoral process. Additionally, an extrajudicial channel was set up in partnership with WhatsApp to report bulk messaging. Furthermore, WhatsApp implemented changes to its infrastructure, restricting the number of contacts to which users can forward messages. This measure aimed to slow down the spread of misinformation and maintain WhatsApp for personal use. Facebook, on the other hand, worked on adhering to the Brazilian General Data Protection Law. This included seeking permission from Brazilian users for specific data usage, updating privacy notices, and providing tools to enhance transparency and control over personal information. Nevertheless, as we will see next, this was not enough to curb influence campaigns in the 2022 election and the organization of anti-democratic movements on Meta’s social media platforms.

In the lead-up to the 2022 elections, WhatsApp experienced a decline in popularity in Brazil, with users shifting towards Telegram. Some of the company's efforts to curb bulk messaging and the dissemination of misinformation, such as limiting the number of recipients for forwarded messages, may have contributed to this shift. In contrast, Instagram gained space as a vector of political participation and Facebook maintained its popularity in the country. In an attempt to prevent a recurrence of issues seen in 2018, Meta took several measures, including: the commitment to establish an Elections Operations Center with a specific focus on Brazil, introducing labels to flag misinformation with links to fact-checking agencies, and implementing tags for electoral material. However, this was not enough to curb the spread of misinformation and attacks on Brazilian democracy.

In the 2022 presidential elections, influence campaigns on Facebook and Instagram were predominantly carried out through ads and boosted posts. An ad is paid upfront whereas a boosted post is first published as a regular post and then the creator pays for the post to be boosted to gain visibility. Global Witness demonstrated a significant flaw in Facebook's ability to detect and combat election-related disinformation in the run-up to the Brazilian election: The civil society organization tested Facebook’s moderation strategy by submitting several ads containing false information on the election or content undermining the legitimacy of the electoral process. The platform’s measures to curb political influence campaigns did not work and the ads were approved by Facebook.

This critical failure underscores the platform's shortcomings in effectively identifying and countering disinformation ahead of crucial political events.

The submitted ads included misleading details about voting procedures, polling locations, and methods, thereby violating Meta’s election ad policies. An investigation by the Brazilian investigative journalism agency Agência Pública also showed how boosted posts were used by several actors, including political candidates, to spread insecurity in the Brazilian electoral process, attacking the voting ballots, and even creating illegal polls. This process sheds light on Facebook’s contradictory attempts in curbing influence campaigns. Despite its aforementioned efforts, the platform’s affordances, such as its ads and boosted posts, enabled the execution of influence campaigns that violated Facebook’s guidelines.

Access to easy tools to share information on platforms represents a notable shift from the 2018 to the 2022 electoral process. In 2018, political groups worked with private companies to influence the process through WhatsApp bulk messaging. However, in 2022, the landscape changed with the rise of Telegram as the go-to app for political organization and with Instagram’s and Facebook's ads and boosted posts, enabling any user to amplify content and create advertisements. Essentially, this platform dynamic created a situation where individuals no longer require a private company as an intermediary to share information with users and influence the political process on a mass scale; Meta itself provides users with all the essential tools to do so. Before, during and after the electoral process, several civil society organizations, such as the Digital Rights Coalition, called out the role of social media, including Meta’s platforms, in skewing the presidential process. Yet, as argued by Lori Regattieri, the tech companies ignored the writing on the wall and the use of Meta ads and boosted posts to disseminate misinformation and undermine democracy continued even after the conclusion of the election process.

Following the election on October 30th, Bolsonaro supporters, unwilling to accept defeat, initiated blockades on major highways nationwide. These blockades later transformed into protests in numerous cities, where supporters established camps in front of military bases that persisted for weeks. This time, social media posts by Bolsonaro supporters called for a military intervention, alleging electoral fraud akin to the "Stop the Steal" movement in the U.S., and urged the public to encamp in front of military bases throughout Brazil, advocating for a coup. An intriguing aspect of this phenomenon is that company profiles with no followers and no products for sale utilized Meta ads – both on Facebook and Instagram – to promote calls for the camps in front of military bases and spread misinformation about the supposed "election fraud" or "stolen elections". In this period, searches in Portuguese for election-related terms such as "fraud," "intervention," and "ballots" on Facebook and Instagram predominantly led individuals to Facebook groups and social media posts casting doubt on the fairness of the elections or overtly advocating for a military coup.

Read more about diffuse actors in our case study, 'Regulating Diffuse Actors in the 2024 Indian Elections' by Amber Sinha.

These calls for action gained traction in the first week of January 2023. Via WhatsApp groups, various entities coordinated a large-scale demonstration in Brazil's capital, involving the transportation of supporters from other states to Brasilia. Examining Meta's platforms, we can say that misinformation and calls to attack the Brazilian democratic institutions were present in all of Meta’s social media. However, discussions about the actual organization of caravans to Brasilia and logistical details of the coup attempt were predominantly held on WhatsApp. Amid media coverage exposing anti-democratic groups on Meta's platforms and highlighting the potential for a coup attempt, both the company and the Brazilian government at that time apparently took no proactive measures to address or curb these issues.

On the day of January 8th, Meta continued to play a significant role. As the protesters invaded government buildings in Brasilia, numerous Instagram Lives showcased the unfolding chaos. Concurrently, false information circulated on Facebook and WhatsApp, claiming that due to the popular mobilization, the army had decided to intervene. This generated an atmosphere of instability, both in terms of information and democratic norms. Towards the day's end, police forces took action and dispersed the protesters. Subsequent days witnessed a series of arrests. On January 9th 2023, Meta stated it was "removing posts from the platform that praise or support anti-democratic demonstrators in Brazil who over the weekend stormed the country's Congress, Supreme Court and presidential palace.”

As of 2024, investigations into the events of January 8th are ongoing. Amidst the detainment of individuals involved in the attacks, a Mixed Parliamentary Inquiry Committee on the Events of January 8 was established. The committee has reached a conclusion highlighting the significant role played by social media as a catalyst and facilitator of the events that transpired on that day. Following ongoing police investigations into the riots and those involved, some individuals have been convicted by the Supreme Court on charges including attempted coup and crimes against the democracy. Bolsonaro is currently under investigation by the Supreme Court for his alleged role in orchestrating the riots, which he denies. Additionally, Brazil's electoral authority has banned Bolsonaro from running for elected office for eight years due to accusations of abusing power and misusing the media in the lead-up to the 2022 election.

This serves as a vivid illustration of the profound impact and influence that private companies can have on real-world events.

The intertwining narratives of Brazilian democracy, anti-democratic riots, and Meta's platforms tell a cautionary tale of the digital age's impact on governance and civic life. The events of January 8th, 2023, marked a disturbing moment where the tools of social media were weaponized to undermine the very foundations of democracy. Despite Meta's efforts to mitigate misinformation and influence campaigns, the platforms it oversees became battlegrounds for political manipulation and extremism. The evolving relationship between Meta and Brazilian politics is shifting: the 2018 elections was characterized by bulk messaging in WhatsApp, and the 2022 electoral cycle was marked by ads and boosted posts. This relationship underscores the layered connections between platform’s affordances and moderation strategies, influence campaigns spearheaded by companies exploiting infrastructural vulnerabilities, the threats posed by misinformation and disinformation and a much needed awareness of safeguarding democratic institutions against digital threats. As investigations continue and legal actions unfold, Brazil stands at a crossroads, grappling with the consequences of unchecked online influence and the imperative to uphold democratic norms in the face of emerging challenges.

---

About the Author:

Pedro Maia works at the intersection of Political Science, International Relations and Science and Technology Studies. He has two main research agendas: internet infrastructures and democracy and the role of technology in security organisations. He is currently a research assistant at the Geneva Graduate Institute’s Tech Hub. Prior to that, he collaborated with different research and civil society organisations working on surveillance, democratic uses of technology, security technologies, GIS and surveillance. Connect with Pedro on X (formerly Twitter) at @pdsmaia, on Blue Sky @pdsmaia.bsk.social or on LinkedIn @Pedro Maia.

---

If you want to find out more about the firms that support political parties engage with Google advertising, head over to The Influence Industry Explorer.

The Influence Industry Project is led since 2016 by Tactical Tech’s Data and Politics team addressing the pervasive data-driven technologies used by political groups within elections and political campaigns.

First published: April 18, 2024