Download PDF

This Essay presents a normative structure for advocating for international soft law standards that can help domestic jurisdictions provide content moderation for election misinformation. Relying on a comparison between the cases of Brazil and the U.S. (both facing recent democratic erosion), this Essay shows how Brazilian courts responded to challenges to democracy and how, in the U.S., content moderation generally depends on private actors. The theoretical analysis presented indicates that transnational and constitutional approaches are required both in the face of the de-territorial characteristics of social media disinformation and also as a prerequisite to conceiving a legitimate approach to private content moderation. This Essay argues that: jurisdictional contextual features cannot be ignored; basic regulations are desirable; content moderation cannot be solely left to private actors, especially considering the need for the protection of democracy; and, finally, that private moderation must also be democratized.


I. Introduction

The digital age allows for instant communication with immediate consequences due to the exercise of free speech around the globe. It also generates an imbalance of economic power well beyond the limits and designs of 20th century capitalism.1 At the same time, two challenges have arisen at the international and domestic levels. One is related to the debate on the feasibility and adequacy of content regulation, especially in situations involving hate speech, influence operations, disinformation on social media, severe human rights violations, and the preservation of democratic institutions. The other challenge involves the boundaries and legitimacy of digital platforms’ deep concentration of economic power, which comes with political and constitutional consequences.

International institutions, however, have consistently lagged behind domestic ones and are still trailing digital regulation. To keep pace with the digital revolution, parliaments, courts, and administrative authorities within national jurisdictions have been pressed to (re)actively respond to concrete threats to both fundamental rights and democratic institutions by users of the internet and digital platforms. On a global scale—except for the European Union (EU) quasi-federal approach to digital markets,2 digital services,3 and personal data protection4 —there is almost no relevant clear international instrument for protecting online users against both hate speech and deliberate aggressions directed toward democratic institutions.

Focusing on the protection of constitutional democracy, this Essay discusses how international and constitutional frameworks can interact with each other to produce normative regulation to respond to the sequential and systematic attacks generated in the digital sphere. Departing from the flat definition of liberal democracy introduced by Tom Ginsburg and Aziz Z. Huq5 —including free and fair elections, rights of speech and association, and the rule of law—we aim to present guidelines for a normative framework that can inform lawmaking bodies, courts, and administrative agencies. These guidelines identify why content moderation is necessary to protect constitutional democracy when it is affected during elections.

The focus of this Essay is the regulation of discourse (speech regulation) during elections. First, this Essay considers the cases of Brazil and the U.S. (both having faced recent democratic decline)6 and conducts a comparative analysis of how these two distinct jurisdictions offer examples of online content moderation during elections. This Essay considers legislative measures, executive policies, and court rulings.

Second, using literature and material on the online electoral context, this Essay extracts the main regulatory principles that can overcome territorial boundaries and address harms to liberal constitutional democracy. These principles should form the foundation of international digital regulation envisaging an overarching goal of protecting democracy online. This framework focuses on non-saturated norms, directives needing further interpretation, and densification via constitutional and statutory legislation, public policies, and judicial rulings inside each jurisdiction that may refer to them.

The cases of Brazil and the U.S. presented in this Essay show that content regulation during elections is especially relevant during times of democratic erosion. The spread of disinformation on digital platforms is a de-territorialized subject that requires standards and practices that can be part of international soft law.

II. Transnationalizing Constitutional Systems

In the 2010s, various scholars devoted their work to understanding how constitutions deal with, absorb, or even reject international law and foreign norms. Their analyses showed that domestic courts play a prominent role in rejecting the influence from foreign and supranational regulations as well as case law.7 Jackson’s proposed trichotomy demonstrated that an engagement model, whereby courts have a more active role in dealing with international and foreign norms, was more appropriate than just opposing international norms (the resistance model) or simply being automatically bounded by such norms (the convergence model).8 In other words, courts are actively reflecting upon their roles in incorporating international and foreign norms and defining how to adapt them domestically. In addition, an obligation to at least consider international material may set up tribunals with a more constructive relationship with supranational and foreign law. For example, the Brazilian Federal Supreme Court cited the American CLOUD Act in order to legitimize domestic law that authorized judges to require data from providers located abroad, much in the style of the engagement model.9

The construction of global norms depends on continuous citing, reflecting upon, and even rejecting supranational, regional, and comparative norms and case law. Kathryn Sikkink proposed one of the most remarkable metaphors illustrating this point. She offered the idea of a “justice cascade” for accountability for crimes against humanity, which followed a constant rhythm throughout the 1990s and 2000s, especially in post-dictatorship Latin American countries.10 Moreover, Bollinger & Callamard argue that, in the field of freedom of expression, one must also consider cross-regional and intra-regional exchanges that help build transnational law.11 These reflections indicate that specific cases debated within domestic jurisdictions may offer normative elements for the construction of international standards in the field of harmful election disinformation spread on internet platforms.

Reflecting upon the role of constitutions, Gunther Teubner showed that the traditional nation-state archetype is no longer sufficient as a paradigm for the different normative spheres that flowed from globalization.12 Instead of isolated domestic juridical systems, what surfaced was the development of a more complex constitutional process in which private and hybrid actors contribute to norm construction. The idea of societal constitutionalism, a form of social organization that demands collegial forms of decision-making in civil society,13 is the starting point for avoiding non-democratic governance. Teubner adds to Sciulli’s contribution the idea that constitutionalism and governance through social, political, and administrative politics will reach private sectors, allow for the autonomy of social sub-constitutions, and create civic rights related to communicative mediums.14 All these constitutional requirements must be followed by “Big Tech” companies if they are to reliably protect constitutional democracies. In other words, private organizations must also legitimize their own forms of content moderation through constitutional mandates.

Against this theoretical background, one must consider the relationship between free and fair elections, the erosion of democracies at a transnational level, and the regulation of free speech against both state and private organizations. Ultimately, this Part argues that constitutionalism, as a form of government for autonomous orders, can both guide transnational orders that include public and private actors and serve as an archetype for the legitimacy of self-governing content moderation inside private platforms.

III. Content Moderation in Brazil: Courts in the Digital Sphere

Contrary to the traditional format of commissions regulating electoral processes, Brazil—at least from the 1930s on—has used tribunals to oversee elections.15 After the 1988 Constitution and the consolidation of democracy that followed two dictatorships (1937–1945 and 1964–1985), organs of the electoral system gained independence for de facto control of free elections in Brazil.16 The main tribunal of the Brazilian electoral judiciary is the Tribunal Superior Eleitoral (Superior Electoral Court)—a hybrid entity with a variety of functions. Brazilian electoral scholar, José Jairo Gomes, describes these functions as: (a) administrative, when it organizes the electoral process; (b) judicial, when it rules upon cases involving the application of the Brazilian electoral statutes; (c) normative, when it enacts resolutions to regulate the application of the electoral statutory system; and (d) advisory, when it is consulted by public authorities and political parties on electoral matters.17 One can easily see both the difficulties and advantages that arise from the complex system in which tribunals both adjudicate conflicts between parties and regulate (as quasi-legislators) how elections take place.

From the 1988 Constitution on, electoral tribunals have been ruling and administering elections without many challenges. The 2018 election, which brought Jair Bolsonaro to the presidency, became the biggest hurdle to the Brazilian electoral supervision system to date. Even during the 2018 presidential campaign, Bolsonaro criticized Brazilian electronic ballots and how the electoral courts administered them.18 In response, the Superior Electoral Court ruled that Google Brasil and Facebook Serviços Online do Brasil should delete videos in which Bolsonaro accused the electronic ballots system of fraud.19 The criticism was viewed as institutionally directed against the electoral courts. One of the dissenting justices defended Bolsonaro’s freedom of expression, including the attacks on electronic ballots.20

Following the authoritarian leader’s playbook, Bolsonaro spent a large part of his term criticizing the electoral courts21 and attacking the electronic ballot system.22 In 2021, his allies in the National Congress tried to approve a constitutional amendment that would return to printed ballots as the exclusive form of voting,23 even after more than twenty years without any significant issues delegitimizing the electronic system. The proposal did not garner enough votes to amend the 1988 Constitution.24 Yet, Bolsonaro continued attacking electoral authorities. He primarily aimed at the three Supremo Tribunal Federal (Federal Supreme Court) justices who rotated into the office of Chief Justice of the Superior Electoral Court in the 2021–2022 period: Justices Roberto Barroso, Edson Fachin, and Alexandre de Moraes.25 Some of the attacks were not only based on the suspicion of fraud but were personal attacks against the justices and their relatives.26

A. Courts Protecting Elections

On one side, courts’ responses consisted of rulings that responded to the attacks on electronic ballots. For example, the Superior Electoral Court initiated a procedure to investigate Bolsonaro for attacking the electronic voting system.27 Such an investigation could have rendered him unfit for the 2022 elections. But the court was not as swift as the situation merited; it should have reached its result before or during the elections.28 In addition, the Superior Electoral Court condemned Bolsonaro’s 2018 campaign for misusing WhatsApp and harming electoral competition, although the ruling had a much more maxi-minimalist approach.29 The court refrained from disqualifying Bolsonaro’s presidential mandate but made a lengthy, strong statement against fake news on elections, advising candidates that the tribunal would be harsher in 2022.

The Superior Electoral Court also used its administrative and normative functions to tackle fake news during elections. Former Superior Electoral Court Chief Justice Roberto Barroso collaborated with civil society organizations and Big Tech representatives to improve the court’s tools to deal with misinformation in electoral campaigns. The tribunal’s lack of action during the 2018 elections showed that it had a lot of work to do to prevent further inaction in the future.30 The Superior Electoral Court created a Program to Tackle Disinformation31 and made more than 55 agreements with state and civil society entities to deal with disinformation.32 The court created a webpage for fact-checking and also developed a chatbot for answering doubts concerning the electoral process.33 The Superior Electoral Court launched digital campaigns and canceled accounts spreading misinformation.34 These actions showed the court taking on a larger role with respect to its administrative functions.35

Particularly during the 2022 presidential elections, civil society organizations and opposition political parties demanded that the Superior Electoral Court act more severely.36 The context of misinformation was larger and established before the election. Consider that, although the Superior Electoral Court did not sanction Bolsonaro with the loss of the presidency for misusing WhatsApp, it condemned a state legislator and removed him from office for unfounded attacks on electronic ballots.37 The court declared that the popular sovereignty of voters to freely elect a representative was truly important.38 The liberty of elector choice, however, could not outweigh abuse of the office and the communication media by legislators, the court declared.39 The Superior Electoral Court’s ruling to remove a legislator from office40 also has a legal basis in Article 22 of Complementary Act 64 of 1990, which protects the normality and legitimacy of elections against the misuse of social communication media.41

B. The Federal Supreme Court Steps In

In 2019, the Federal Supreme Court continued protecting the integrity of elections by inaugurating a procedure to investigate internet attacks against itself and its justices—the so-called “fake news inquiry.”42 The rapporteur of the inquiry was Justice Alexandre de Moraes. It is a broad investigation without a clear deadline, and has a suspect procedure by which the investigator doubles as the adjudicator. The Federal Supreme Court, nonetheless, recognized the inquiry as constitutional since it was based on internal procedures aimed at protecting democracy.43 Freedom of expression could not include hate speech or public menace, according to the court.44

The fake news inquiry included a huge number of facts and investigated persons. Those investigated were Brazilian citizens with a certain number of followers who seriously and recurrently attacked either the Federal Supreme Court or its members on social media. Moreover, the outrageous campaign that Bolsonaro and his supporters launched against Brazilian constitutional democracy resulted in other inquiries. Because the Federal Prosecution Office was already captured by Bolsonaro,45 those investigations were much less prominent than the fake news inquiry. The Federal General Prosecutor had halted more than 100 investigations against Bolsonaro,46 so he would not also support deep investigations in the inquiries he started to purportedly protect democracy. Consider also that Brazilian tribunals’ judges have significant monocratic powers permitting provisional decisions. Justice Moraes saw the need to curtail the threats to democracy in several cases, drawing the attention of Bolsonarism and the media to his decisions.

In addition, in 2022, Justice Moraes became the Superior Electoral Court’s Chief Justice.47 Under his leadership, the court adopted a resolution, based on its normative powers, to control disinformation that could taint the electoral process.48 The resolution repeated statutory provisions from the Brazilian Electoral Code (especially Article 323)49 that forbid the spread of known false facts that could harm the procedures of voting, vote counting, and vote totalization.50 Digital platforms must immediately remove the false content, or face fines ranging from $20,000 to $30,000 per hour of noncompliance.51 Through the indication of the appropriate URLs, the Chief Justice can also demand the removal of identical content shared on other websites and platforms.52 The resolution authorizes the suspension of profiles, accounts, and channels that systematically produce disinformation, as well as the suspension of general access to the platform’s services in cases in which the company is reluctant to obey a judicial warrant.53

The harsh tone of the resolution was not enough to prevent Bolsonaro’s supporters from discrediting the election of President Luiz Inácio Lula da Silva. Nor did it avoid the coup attempt of January 8, 2023, when Bolsonarists invaded the buildings of the National Congress, the Presidency, and the Federal Supreme Court to destroy them.

Although Justice Moraes’s performance attracted criticism in Brazil and abroad, most of his rulings were insufficient to contain the Bolsonarist rage. At the end of 2022, Federal Supreme Court Chief Justice Rosa Weber pressed the full bench to approve a resolution demanding that every ruling by a single justice be immediately submitted to the panels of the court.54 This was the case for several of Justice Moraes’s rulings. The Federal Supreme Court panels agreed with Justice Moraes that energic measures should be taken to protect Brazilian democracy when those rulings were submitted to approval of the other justices.55 Consider also that the Court held that Brazilian judges could require data from internet providers located abroad, both via diplomatic instruments or directly from a company based in Brazil or even in a foreign country.56 The ruling was adopted in a constitutional review procedure by the majority of the members of the court, with only the two justices nominated by Bolsonaro in the minority.

The opinion provided by Justice Gilmar Mendes in the abovementioned case declared that there was a movement of re-territorialization of the internet, especially concerning cybercrimes.57 To give one example, the Brazilian Civil Framework of the Internet58 follows Article 18 of the Budapest Convention, which demands the protection of data collected inside Brazilian territory.59 In any case, one should bear in mind the exterritoriality of internet data. The Court held that provisions of the Brazilian Civil Procedure Code, the Brazilian Criminal Procedure Code, and the Brazilian Civil Framework of the Internet allowed for direct judicial requisition of data, even through diplomacy.60 The Court, however, did not limit itself to those decisions, urging the legislature to build an enhanced legislative basis through statutory law that could allow for more effective data requests between domestic and foreign courts.61

So, at least in the field of free speech on the internet, Brazil appears as a particular example of the demand for regulation from the state and the limits of private self-control. Brazilian high courts were pressed by a polarized context in which direct attacks on democratic procedures demanded quick and effective rulings that regulated, at least at a baseline, free speech on the internet.

Furthermore, the January 8, 2023, coup attempt was also carried out by potential criminal organizations funded by private entities and persons who used social media and messaging apps to coordinate their actions. The attacks were articulated and incubated by employing the interaction between online extremist users and the unity behind the planning and financing forces of the attacks.62 In a certain fashion, the evidence presented so far legitimizes the claim that free speech online and domestic electoral systems became even more core issues in designing democratic-driven models for content governance online, either by state regulators or by social media platforms. In this regard, domestic experiences could be exported to the international level for broader scrutiny by states, international organizations, and other non-state actors. Finally, the solutions envisaged for the interplay between content moderation and elections may be confronted with the ultimate need to promote models of democratic self-defense.63

The Brazilian case indicates that, although there is plenty of legislation regulating fundamental rights on the internet and data protection,64 including on the constitutional level,65 and regulations for a few aspects of electoral campaigning on digital grounds,66 there is still a need for more incisive action by the judiciary branch. The intensity of Bolsonaro’s attacks on Brazilian democracy demanded harsh measures: more than a normative structure, there was the need for judicial rulings protecting the electronic voting system against disinformation, as well as direct measures to halt the coup attempt that was organized via messaging groups. However, these measures can stress courts’ legitimacy by putting an excessive burden on them.

The Brazilian case indicates that content moderation in elections is not only dependent on state regulations. The participatory procedures of the Electoral Superior Court showed that elements of societal constitutionalism are important to provide effective and legitimate tools for controlling information in the digital sphere.

IV. The United States Case: Freedom of Speech vs. Content Moderation?

Donald Trump’s presidency was replete with attacks on American electoral institutions. He created the Presidential Advisory Commission on Election Integrity, a commission with the purported objective of finding vulnerabilities in the voting system.67 The commission was controversial, facing several lawsuits from states and its own members until Trump terminated it.68 President Trump, like Bolsonaro, attacked the ballots beginning in 2016,69 and especially after his defeat in 2020.70 His well-known distortion of the term “fake news” was important in his attacks on traditional media, creating a recipe that Bolsonaro copied in Brazil.

While the U.S. does not totally lack public discourse restrictions, First Amendment protections are so strong that meaningful control of disinformation in elections is a work in progress. The U.S. Supreme Court’s freedom of speech doctrine sets a severe standard that would likely apply to controlling disinformation on the internet during elections—most speech is protected unless it is directed at or is likely to produce imminent unlawful action.71 In the face of U.S. v. Alvarez, even false claims may have value in public discourse,72 a sign that the Supreme Court would probably demand very strong arguments for allowing legislation that controls false content on social media.

First Amendment doctrines, up to now, have influenced the scarce amount of federal legislation concerning the regulation of content moderation. David Ardia, Evan Ringel, and Allysan Scatterday found that there is federal and state legislation concerning protection from false claims against election procedures.73 In 2021, 38 states had legislation of this kind.74 Gielow Jacobs points out that the Supreme Court’s case law on freedom of speech demands very narrow tailoring for speech regulations to be enforceable.75 In Jacobs’s view, the state statutes did not meet that rigorous standard.76 For example, Tompros, Crudo, Pfeiffer, and Boghosian discuss a New York statute that criminalizes falsely reporting an incident without an intent to cause harm.77 The legislation, in this sense, would not survive the Alvarez test.78

It is also worth noting that the Clarifying Lawful Overseas Use of Data Act (CLOUD Act) allows judicial warrants to reach internet providers’ data, including data stored outside American territory.79 The CLOUD Act affects online freedom of speech as it requires that foreign governments’ orders to access data in the U.S. do not infringe freedom of speech, though nothing is said about what version of freedom of speech is protected.80 The CLOUD Act excuses companies from complying with judicial orders in cases in which the user is not American or does not reside in the U.S. and in cases in which there is a chance of violating foreign law with the disclosure of the data.81 The CLOUD Act, in facilitating the delivery of data by internet providers operating in the U.S., does not exempt the legislation from criticism. There is much debate about the harmful effects that CLOUD Act-related agreements could have on privacy and human rights. To some extent, enforcement of the CLOUD Act may undermine the privacy rights of individuals both inside and outside the U.S. and would support a relaxed procedural approach for foreign governments to access internet providers in the U.S. without judicial review or a sufficient degree of transparency.82

In any case, the CLOUD Act could provide a basis for content moderation by judicial authorities in the U.S. when the protection of democracy during elections is at stake. As mentioned above, in holding that it is constitutional to require data from internet providers located abroad, the Brazilian Federal Supreme Court expressly cited the American CLOUD Act to defend its more interventionist approach to data privacy.83 The position of the Brazilian court shows a kind of engagement approach to foreign law and is an example of how online content moderation depends on the transnational flux of norms and regulations.84 The existence of soft law parameters for cases like this one would be helpful to prevent content moderation from violating data protection rights and free speech.

During the 2020 elections in the U.S., much was expected of the centrality of social media content moderation. This was especially true in the context of the intense debate on the capacity of large platforms to create trustworthy systems and to alternatively design self-regulatory patterns for content moderation in democracies.85 However, self-regulatory frameworks for content moderation have already proven insufficient to tackle the complexities of online disinformation and extremist content, as well as the influence operations initiated by domestic and external actors during elections. In 2020, Facebook tried to counter election misinformation targeting Tunisia, Togo, Côte d’Ivoire, and seven other African countries in a joint strategy that culminated with the erroneous removal of accounts owned by Tunisian journalists, activists, and NGO representatives.86 Such accounts had been quite proactive, for instance, in using social media tools to engage in political content during the country’s 2011 revolution.87

In 2020, Twitter blocked a New York Post article about now-President Joe Biden’s son because it was based on hacked materials, then overturned the block two days later.88 After establishing limits on political advertising in early September 2020, Facebook promised not to further amend its policies ahead of the elections.89 Three weeks later, the platform announced changes to its political advertising policies, then blocked a range of speech-related content it had previously allowed.90 In all these cases, users actively sharing, imparting, and accessing political information are left behind and increasingly mistrust the moderation choices about politics and what might be blocked next.

There might be some progress ahead, but social media platforms alone cannot control the presence of several non-state entities or foreign states influencing the U.S. electorate. Although technology is in the hands of these platforms, there are security and diplomatic issues that require governmental policies. As stated by the Select Committee to Investigate the January 6, 2021 attack on the U.S. Capitol, the rise of influence operations consolidated the projection of power exercised by foreign entities over the U.S. audience directly, as a way to “engage . . . in disguised efforts to influence U.S. public opinion.”91 A symbiotic approach is taken regarding the use of social media since influencers benefit from algorithms designed by platforms bringing “congenial messages and other information to users whose views are likely to be similar or compatible.”92 Even for U.S. regulators, however, there is a risk of duplicated efforts in constraining the use of social media. This is the case where masking tools are applied by foreign influencers to amplify U.S.-originated messages targeting a broader domestic audience. Insofar as unilateral governmental control is directed at U.S. users, it would also pose risks to individual rights, including the right to privacy and freedom of speech.

The U.S. case indicates that broader protection of free speech limits the way in which institutions deal with disinformation in the context of elections. As opposed to Brazilian courts, American courts are far from deciding upon the limitation of profiles, engaging in direct content moderation, and even controlling how private platforms manage disinformation. Although the protection of free speech is vital for democracy, online disinformation can affect the way electors make their decisions. Content moderation has a fundamental role in the digital sphere and non-regulation does not seem to be an option.

The next section will address the relationship between the health of democracy, free speech in elections, and digital media, especially considering international law.

V. Free and Fair Elections: Free and Fair Speech in the Digital Age

Besides respect for the fundamental rights of speech and association and the predictability of the rule of law, Ginsburg and Huq argue that free and fair elections, with voting access and rotation of power, are essential to preserving the core of constitutional liberal democracy.93 When these features are in peril, a process of democratic erosion has already likely begun. The digital age, however, distorted the supposedly clear definitions of some of the elements Ginsburg and Huq indicate. Free speech and fair elections enter a blurry relationship when unregulated content is spread during an electoral process that happens with no state oversight to remake the balance between candidates and political parties affected by the abuse of communication and economic power. That is why authorities should consider not only assuring free and fair elections but also free and fair speech when it comes to internet media. That seems to be what Brazilian courts learned from the 2018 presidential elections and tried to apply in 2022.

Even considering freedom of expression as a keystone of constitutional democracies, countries all over the world regulate communication in elections. For both candidates and political parties, the digital age provided states with accessible tools to reach and convince electors from very different segments of the population. A study concentrating on Europe showed how the relationship between the internet and electoral campaigns brought out questions related to: (a) comparisons to broadcasting regulation and its limits concerning digital media; (b) how to deal fairly with all candidates and political parties’ online expenditures; (c) methods for avoiding targeting specific groups without imparting their autonomous decisions and excluding electors; (d) the role of Big Tech companies and algorithmic manipulation; (e) how to deal with misinformation and disinformation; (f) how to measure public opinion online, beyond the traditional methods adopted for polls; and (g) what to do to assure transparency on public internet spending.94

International law instruments do not encompass the internet and electoral problems raised in the European context by the abovementioned study. Normative provisions for free and fair elections can be seen in different international law regulations, none of them dedicated to the digital sphere. The International Covenant on Civil and Political Rights (ICCPR) provides that elections shall be genuinely periodic and guided by universal and equal suffrage, with the protection of the free expression of electors.95 The United Nations Convention Against Corruption (UNCAC) establishes a goal of transparency in the funding of candidates and political parties.96 The American Convention on Human Rights (ACHR) uses a version close to the one adopted by the ICCPR;97 Article 3 of the European Convention on Human Rights (ECHR) follows the same pattern.98

Article 19 of the Universal Declaration of Human Rights99 guarantees freedom of expression and provides that a right to opinion and expression shall be protected against interference; receiving and imparting information shall be allowed to happen through any media, regardless of borders. The cases of Brazil and the U.S. show that these rights can be abused, affecting democratic elections. The ECHR provides a similar formulation of the right to freedom of expression in Article 10(1).100 Article 10(2) limits freedom of expression in national security, territorial integrity, and public safety contexts.101 Article 10(2) also limits freedom of expression for purposes of fighting criminal disorder, protecting health and morals, defending the reputation or rights of others, preventing the disclosure of information received in confidence, and maintaining the authority and impartiality of the judiciary.102 Article 13 of the ACHR protects freedom of expression, foreseeing other means of diffusion of information.103 However, it forbids censorship at the same time that it imposes liability for harming another’s reputation and rights, as well as affirming some of the values that appear in the ECHR.104

The ACHR has a more detailed provision on limiting freedom of expression. It protects that right against the abuse of private control of any medium for spreading information.105 It also excludes hate speech and lawless violence against any person or group of persons from freedom of expression protections.106 In this sense, the ACHR allows for more limitations on expression than its universal and regional counterparts. In this sense, the ACHR can act as a helpful normative tool for Brazilian courts in limiting disinformation that can impair competitive elections and erode democracy.

A normative basis for international law regulation of digital media that has implications for internet international law is Article 19 of the ICCPR. Using both constructivist and originalist interpretations, Molly Land argues that the provision, when mentioning “media,” was supposed to include new technologies developed subsequently.107 Moreover, the debates that led to Article 19’s adoption included private actors, and did not include requests for a state action doctrine.108 The consequences of this reading affect both internet access as a means of expression and how technological design will impact rights. As Land opposes direct enforcement of Article 19 of the ICCPR by courts, it is hard to see how Big Tech will abide by the provision without enforcement by tribunals.109 The results of elections in the past decade have shown that these companies, most of them animated by the defense of the complete absence of norms, use the protection of freedom of expression as an excuse to interfere in cases that affect fundamental rights and democracy.

The European, American, and Brazilian contexts have indicated diverse, problematic consequences of using the internet for abusive purposes during elections. The international normative architecture offers basic standards that help domestic interpretations of how to deal with content moderation. However, the absence of more direct provisions, even in the form of soft law, creates new opportunities for violating fair competition in elections, facilitating the erosion of democracy.

VI. Conclusions: Lessons from the Domestic to International Level

Transforming speech into action is not a privilege of older forms of political organization. Modernization has brought not only the instant worldwide dissemination of information, but also concentrated power. It has enabled suspect sources of disinformation that direct people to confirm their most primitive biases against pluralism and democracy. January 6, 2020, in the U.S. and January 8, 2023, in Brazil demonstrated how that erosion can turn into collapse and how words can become violence. Digital media in particular has catalyzed and accelerated such changes.

International law has a special role in influencing how digital trafficking of information in a de-territorialized internet must be regulated. The normative framework for the digital sphere asks for transnational regulation in terms of societal constitutionalism: state and private organizations must democratically participate in the construction and application of norms that can help fight disinformation in electoral contexts. Concerned specifically with protecting democracy during elections, this Essay posits the following normative directives that flow from each domestic constitutional experience:

  1. Domestic context must be respected. There are very different free speech traditions that should be preserved. The level of content regulated is context-dependent, particularly when it comes to protecting democracy in electoral periods. The Brazilian and U.S. cases indicate contexts in which the control of free speech is very different; however, both of them indicate that some basic level of concern with protecting democratic institutions in elections is appropriate. The contexts indicate that discussions regarding content moderation must consider how specific jurisdictions will circumscribe standards, even if they are based on international law.
  2. No regulation at all is not a solution. States must regulate—according to their constitutional traditions—the level of disinformation produced during electoral periods. Radical attacks that publicly defend violence against democratic institutions become effective actions and should be controlled by both the state and the platforms’ content governance. Even in the constitutional context of the liberal case law of the U.S. Supreme Court, violence should not be tolerated as a form of free speech. The Brazilian case shows that courts’ rulings are the last resort when both state agency regulation and private self-moderation fail. This type of normative parameter should guide content moderation on digital media.
  3. Alone, digital platforms’ self-governance of content moderation is not enough. State organs with civil society participation (involving users, Big Tech, experts, and NGOs) will have the legitimacy to require responses from digital platforms. When those platforms fail to respond, courts (along with civil society) should have the power to employ effective remedies. Considering the Brazilian case, digital platforms were slow to respond to judicial warrants, pressing the Superior Electoral Court to adopt a harsher resolution that made it jointly responsible for controlling the spread of misinformation.
  4. Democracy is protected with more democracy. Digital platform self-governance for content moderation must be reinforced by civil society participation. Moderation must count on users’ representation or direct participation. The level of economic power large technology firms have must be balanced with a form of democratic governance, societal constitutional legitimacy, and a desirable, socially responsive regulation. Future domestic regulations in jurisdictions like the U.S. and Brazil must allow for more civil society participation on Big Tech companies’ content oversight boards.

The comparison between the cases of Brazil and the U.S. merits at least two concluding remarks. First, the Brazilian internet regulation framework is closer to the European context. Even with plenty of statutory law, when disinformation harms electoral procedures, judicial rulings control companies’ content moderation. Stressing the role of courts can, however, create problems of legitimacy in the future. Second, U.S. institutions seem to leave content moderation tasks to Big Tech companies to better protect free speech. This can be a risky position, especially when democracy is in danger.

The normative directives that appear in this Essay show that regulation is not incompatible with free speech. They are part of soft law guidelines that help maintain an international public sphere in which domestic jurisdictions exchange experiences without, for now, the need to rely on an international treaty for the protection of democracy against disinformation in election periods.

  • 1Shoshana Zuboff, The Age of Surveillance Capitalism 128 (2019).
  • 2Commission Regulation 2022/1925 of Sept. 14, 2022, On Contestable and Fair Markets in the Digital Sector and Amending Directives (EU) 2019/1937 and (EU) 2020/1828 (Digital Markets Act), 2022 O.J. (L 265).
  • 3Commission Regulation 2022/2065 of Oct. 19, 2022, On a Single Market for Digital Services and Amending Directive 2000/31/EC (Digital Services Act), 2022 O.J. (L 277) [hereinafter DSA].
  • 4Commission Regulation 2016/679 of Apr. 27, 2016, On the Protection of Natural Persons with Regard to the Processing of Personal Data and on the Free Movement of Such Data, and Repealing Directive 95/46/EC (General Data Protection Regulation), 2016 O.J. (L 119); DSA, supra note 3.
  • 5Tom Ginsburg & Aziz Z. Huq, How to Save a Constitutional Democracy 47(2018). As one of us previously argued, the contours of liberal democracy offered by Ginsburg and Huq form the baseline currently under attack in different jurisdictions that face democratic erosion. Emilio Meyer, Constitutional Erosion in Brazil 9 (2018). However, cases such as Brazil, India, México, Colombia, and others presuppose a more substantial definition of constitutional democracy that can also be in danger of erosion. Id. For the purpose of this Essay, their definition is, nonetheless, sufficient because it encompasses the need to protect free and fair elections.
  • 6See generally Ginsburg & Huq, supra note 5; Steven Levitsky & Daniel Ziblatt, How Democracies Die 152 (2018); Robert R. Kaufman & Stephan Haggard, Democratic Decline in the United States: What Can We Learn from Middle-Income Backsliding?, 17 Persps. on Pol. 417 (2019); David E. Pozen & Kim Lane Scheppele, Executive Underreach, in Pandemics and Otherwise, 114 Am. J. Int’l L. 608, (2020); Tom Daly, Understanding Multi-directional Democratic Decay: Lessons from the Rise of Bolsonaro in Brazil, 14 L. & Ethics Hum. Rts. 199 (2020); Meyer, supra note 5. The 2023 V-DEM report indicates the U.S. and Brazil—along with Indonesia—as “autocratizer” countries in the last ten years. V-Dem Institute, Democracy Report 2023: Defiance in the Face of Autocratization 13 (2023), In other words, these jurisdictions have poor indicators concerning liberal democracy’s indexes. See id. at 50.
  • 7See e.g., Vicki C. Jackson, Constitutional Engagement in a Transnational Era 162 (2010); see also Gunther Teubner, Constitutional Fragments: Societal Constitutionalism and Globalization 114 (2012).
  • 8See Jackson, supra note 7, at 283.
  • 9S.T.F., Ação Declaratória de Constitucionalidade No. 51 Distrito Federal, Relator: Ministro Gilmar Mendes, 10.12.2020, 290/2020, Diário da Justiça Eletrônico [D.J.e.], 11/12/2020, 18–19 (Braz.) [hereinafter ADC 51].
  • 10Kathryn Sikkink, The Justice Cascade 120 (2011).
  • 11Lee C. Bollinger & Agnes Callamard, Regardless of Frontiers 1, 9 (2021).
  • 12Teubner, supra note 7, at 91.
  • 13David Sciulli, Theory of Societal Constitutionalism 150 (1991).
  • 14See generally Teubner, supra note 7.
  • 15Tribunal Superior Eleitoral, Practical Guide: 2022 Brazilian Elections 40 (2022),
  • 16Constituição Federal [C.F.] [Constitution] arts. 118–22.
  • 17José Jairo Gomes, Direito Eleitoral 160 (2020).
  • 18See Jack Nicas et. al., How Bolsonaro Built the Myth of Stolen Elections in Brazil, N.Y. Times (Oct. 25, 2022),
  • 19TSE determina exclusão de vídeo em que Jair Bolsonaro critica urnas eletrônicas, Tribunal Superior Eleitoral (Oct. 25, 2018),
  • 20Id.
  • 21See, e.g., Andrew Downie, Bolsonaro’s Attack on Brazil’s Electoral System Sparks Outrage, Guardian (July 19, 2022),
  • 22See, e.g., Anthony Boadle, Bolsonaro Attacks Brazil’s Election System in Briefing for Diplomats, Reuters(July 18, 2022),
  • 23Bolsonaristas tentam aprovar voto impresso na CCJ da Câmara, Congresso Em Foco (July 5, 2021),
  • 24Câmara rejeita proposta que tornava obrigatório o voto impresso, Agência Câmara de Notícias (Aug. 10, 2021),
  • 25Bolsonaro chama Barroso de “sem caráter” e ataca Moraes, Poder360 (June 11, 2022),
  • 26Id.
  • 27Lisandra Paraguassu, Brazil Court to Probe Bolsonaro for Attacks on Voting System, Reuters(Aug. 2, 2021),
  • 28Emilio Peluso Neder Meyer & João Andrade Neto, Courts Are Finally Standing Up to Bolsonaro, Verfassungsblog (Aug. 9, 2021),
  • 29Yvonne Tew, Strategic Judicial Empowerment, Am. J. Comp. L. (forthcoming 2023) (manuscript at 65–67),
  • 30Letícia Casado, TSE falha no combate a fake news na campanha de primeiro turno, Folha de S. Paulo (Oct. 5, 2018),
  • 31Disinformation and Fake News: Program to Confront Disinformation: Elections 2020, Superior Electoral Ct.,
  • 32Disinformation and Fake News: Partner Institutions, Superior Electoral Ct.,
  • 33Disinformation and Fake News: 10 Innovative Initiatives, Superior Electoral Ct.,
  • 34Id.
  • 35Confira as ações contra a desinformação efetivadas pelo TSE nos últimos anos, Tribunal Superior Eleitoral (Aug. 11, 2022),
  • 36Lucas Teixeira, Campanha de Lula quer reunião com TSE para pressionar Telegram, UOL Eleições (May 23, 2022),
  • 37Deputado Francischini é cassado por propagar desinformação contra a urna eletrônica, Tribunal Superior Eleitoral (Oct. 28, 2021),
  • 38Id.
  • 39Id.
  • 40T.S.E., Recurso Ordinário Eleitoral No. 0603975-98, Relator: Ministro Luis Felipe Salomão, 28.10.2021, 228, Diário da Justiça Eletrônico [D.J.e]10/12/2021 (Braz.).
  • 41Lei Complementar No. 64, de 18 de Maio de 1990, Diário Oficial da União [D.O.U.] de 21.05.1990 (Braz.).
  • 42S.T.F.J., INQ 4781 (2023) (Braz.),
  • 43Plenário conclui julgamento sobre validade do inquérito sobre fake news e ataques ao STF, Supremo Tribunal Federal (June 18, 2020),
  • 44Id.
  • 45Rafael Neves, PGR já arquivou 104 pedidos de investigação contra Bolsonaro vindos do STF, UOL (July 30, 2022),
  • 46Id.
  • 47Brazil’s Superior Electoral Court Has New Chief Judge, AgênciaBrasil (Aug. 17, 2022),
  • 48Resolução No. 23.714, de 20 de Outubro de 2022, Diário da Justiça Eletrônico [D.J.e], de 24.10.2022 (Braz.).
  • 49Lei No. 4.737, de 14 de Julho de 1965, Diário Oficial da União [D.O.U.] de 19.07.1965 (Braz.).
  • 50Resolução No. 23.714, supra note 48.
  • 51Id. These figures, expressed in U.S. dollars, are based on a BRL to USD rate of R$5.20 to US$1. SeeCotações e boletins, Banco Central do Brasil,
  • 52Resolução No. 23.714, supra note 48.
  • 53Id.
  • 54Emenda Regimental Altera Regras Para Devolução de Pedidos de Vista no STF, STF (Dec. 26, 2022),
  • 55STF Referenda, Por Unanimidade, Determinação de Desbloqueio de Rodovias, STF (Nov. 1, 2022),; STF Confirma Decisão que Impôs Medidas sobre Manifestação Antidemocrática, STF (Jan. 12, 2023),; STF Respalda Moraes e Mantém Governador do DF Afastado e Prisão de Torres, Folha de S. Paulo (Jan. 11, 2023),
  • 56ADC 51, supra note 9, at 25.
  • 57Id. at 13–14.
  • 58Lei No. 12.965 de 23 de Abril de 2014, Diário Oficial da União [D.O.U.] de 24.04.2014 (Braz.), art. 11 [hereinafter Brazilian Civil Framework of the Internet].
  • 59Convention on Cybercrime, Nov. 23, 2001, E.T.S. No. 185. For support for the claim that the Brazilian Civil Framework of the Internet follows Article 18 of the Budapest Convention, see ADC 51, supra note 9, at 29.
  • 60ADC 51, supra note 9, at 10.
  • 61Id. at 35.
  • 62Caios Matos, Campanha nas redes sociais identifica golpistas bolsonaristas, Congresso em Foco (Jan. 11, 2023),
  • 63See Laura Fichtner, Content Moderation and the Quest for Democratic Legitimacy, 2 Weizenbaum J. Digit. Soc’y 1 (Dec. 31, 2022).
  • 64See, e.g., Brazilian Civil Framework of the Internet, supra note 58; Lei No. 13.709 de 14 de Agosto de 2018, Diário Oficial da União [D.O.U.] de 15.08.2018 (Braz.).
  • 65C.F. art. 5.
  • 66Código Eleitoral [C.E.] [Electoral Code] arts. 326-A, 326-B.
  • 67Levitsky & Ziblatt, supra note 6, at 152.
  • 68Charles Stewart III, Trump’s Controversial Election Integrity Commission is Gone. Here’s What Comes Next., Wash. Post (Jan. 4, 2018),; Matthew Dunlap, I’m on Trump’s Voter Fraud Commission. I’m Suing It to Find Out What It’s Doing., Wash. Post (Nov. 30, 2017),; Background on Trump’s ‘Voter Fraud’ Commission, Brennan Ctr. for Just. (July 18, 2017),
  • 69See, e.g., Darren Samuelsohn et al., Trump Seizes on Isolated Glitches to Fuel ‘Rigged’ Election Claims, Politico (Nov. 8, 2016),
  • 70See, e.g., Tal Axelrod, A Timeline of Donald Trump’s Election Denial Claims, Which Republican Politicians Increasingly Embrace, ABC News(Sept. 8, 2022),
  • 71See, e.g., Brandenburg v. Ohio, 395 U.S. 444 (1969); accord Texas v. Johnson, 491 U.S. 397 (1989); Citizens United v. Fed. Election Comm’n, 558 U.S. 310 (2010).
  • 72567 U.S. 709, 718–22 (2012).
  • 73See David Ardia et al., State Regulation of Election-Related Speech in the U.S.: An Overview and Comparative Analysis 29–30 (UNC Ctr. for Media L. & Pol’y ed., 2021).
  • 74Id. at 3.
  • 75Leslie Gielow Jacobs, Freedom of Speech and Regulation of Fake News, 70 Am. J. Comp. L. 278, 295 (2022).
  • 76Id. at 294.
  • 77Louis W. Tompros et al., The Constitutionality of Criminalizing False Speech Made on Social Networking Sites in a Post-Alvarez World, 31 Harv. J.L. & Tech. 65, 85–86 (2017); N.Y. Penal Law § 240.50.
  • 78See Tompros et al., supra note 77, at 85–86.
  • 79Consolidated Appropriations Act, 2018, Pub. L. No. 115-141, 132 Stat. 1213 (2018) [hereinafter CLOUD Act].
  • 80See Greg Nojeim, Cloud Act Implementation Issues, Lawfare (July 10, 2018),
  • 81CLOUD Act, supra note 79.
  • 82See, e.g., Neema Singh Guliani & Naureen Shah, The CLOUD Act Doesn’t Help Privacy and Human Rights: It Hurts Them, Lawfare (Mar. 16, 2018),; Eugenia Lostri, The CLOUD Act, Ctr. for Strategic & Int’l Stud. (Oct. 2, 2020),
  • 83See supra note 9 and accompanying text.
  • 84Jackson, supra note 7, at 164.
  • 85Marietje Schaake & Rob Reich, Election 2020: Content Moderation and Accountability, Stan. Cyber Pol’y Ctr. (Oct. 20, 2020),
  • 86Simon Speakman Cordall, Facebook Deactivates Accounts of Tunisian Political Bloggers and Activists,Guardian (June 4, 2020),
  • 87Leil Zahra, Facebook Delivers a Serious Blow to Tunisia’s Music Scene, Mozilla Found. (June 16, 2020),
  • 88Kate Conger & Mike Isaac, In Reversal, Twitter Is Not Longer Blocking New York Post Article, N.Y. Times (Apr. 17, 2021),
  • 89Twitter Lifts Freeze from New York Post Account After Policy Reverse, Guardian (Oct. 30, 2020),
  • 90Id.
  • 91H.R. Rep. No. 117-663, at 807 (2022).
  • 92Id. at 810.
  • 93See Ginsburg & Huq, supra note 5, at 10.
  • 94Damian Tambini, Comm. of Experts on Media Pluralism & Transparency of Media Ownership [MSI-MED], Internet and Electoral Campaigns: Study on the Use of Internet in Electoral Campaigns 8 (2018).
  • 95International Covenant on Civil and Political Rights art. 25(b), Dec. 16, 1966, 999 U.N.T.S. 171.
  • 96Convention Against Corruption art. 7.3, Dec. 9, 2003, 2349 U.N.T.S. 41.
  • 97American Convention on Human Rights art. 23(b), Nov. 22, 1969, 1144 U.N.T.S. 123 [hereinafter ACHR]
  • 98Convention for the Protection of Human Rights and Fundamental Freedoms, Nov. 4, 1950, E.T.S. No. 5 [hereinafter ECHR].
  • 99G.A. Res. 217 (III) A, Universal Declaration of Human Rights (Dec. 10, 1948).
  • 100See ECHR art. 10(1).
  • 101See id. art. 10(2).
  • 102Id.
  • 103See ACHR art. 13.
  • 104Id. art. 13(2).
  • 105Id. art. 13(3).
  • 106Id. art. 13(5).
  • 107Molly Land, Toward an International Law of the Internet, 54 Harv. Int’l L.J. 393, 394 (2013).
  • 108Id. at 395.
  • 109Id.