How Brazil’s X ban signals growing control over online free speech

In August 2024, Brazil’s Supreme Court President Alexandre de Moraes imposed a ban on X (formerly Twitter) after the platform ignored previous court orders to remove user profiles. This decision was widely reported in the international press, with headlines suggesting that the ban resulted from X owner Elon Musk’s disregard for Brazilian law, which requires foreign companies to have a local representative. However, the situation is more complex, and the court ruling requiring X to appoint a legal representative in the country cannot be understood in isolation.

Prominent figures such as economist Thomas Piketty and Nobel laureate Daron Acemoglu signed an open letter supporting the X ban, framing it as a necessary stance against “Big Tech” companies and their perceived influence on Brazilian democracy, which they described as an effort toward “digital sovereignty.” However, a closer examination reveals that the real motivation behind the X ban was not related to digital sovereignty, but rather a broader effort by the Brazilian Supreme Court to control political discourse and suppress dissenting voices in the country.

This situation is not unique to X; similar rulings have also resulted in the banning of Rumble and Telegram in the country. In these instances, the Brazilian Supreme Court banned entire platforms for failing to comply with the ban on a single user profile – that of political blogger Allan dos Santos, a Brazilian residing in the US.

While X has since resumed operations in Brazil, another Supreme Court case regarding the constitutionality of Brazil’s Internet Law is underway. This case could impose further restrictions on freedom of speech in the country and raise concerning implications for Brazil’s internet governance.

This should be viewed as a cautionary tale for the United States and all democratic nations regarding what unchecked discretionary power over digital speech can gradually lead to. In times of crisis, discretionary censorship is often normalized and even presented as benevolence rather than repression. Once this power is granted, it tends to expand.

Brazil’s Internet regulations and constitutional protections

This situation began when Judge Alexandre de Moraes ordered X to ban profiles accused of spreading disinformation and inciting “attacks on democracy.” X publicly refused to comply due to concerns about free speech and because they argued that the orders were against Brazilian law. This prompted Moraes to threaten to arrest X’s representative in the country. In response to legal threats against its representative, X closed its Brazilian offices, leaving the company without an official representative, a legal requirement for operating in Brazil. This status eventually became the foundation for the ban on X.

In 2014, the Brazilian Internet Law (Marco Civil da Internet, Law 12.965, Art. 19, §1) was introduced to regulate online content. Article 19 states that platforms can only be held liable for third-party content after receiving a specific court order, and such an order must identify the infringing content, which is equivalent to the U.S. Section 230 of the Communications Decency Act of 1996.

Specifically, it establishes that “internet application providers can only be held civilly liable for damages arising from third-party content if, after a specific court order, they fail to take action.”

Thus, initially, it might seem that X was rightfully found liable, given that it received several “specific court order[s]” requesting to take down accounts and did “fail to take action […] to make the content identified as infringing unavailable.” But a closer examination suggests that X’s legal team was fundamentally correct in challenging these court orders because they did not contain the requirements for censorship established by law.

First, in the court orders ignored by X, Moraes required the platform to ban profiles, but Article 19 only allows the censorship of content. It’s reasonable to argue that being censored for a specific statement should not lead to someone being permanently banned from expressing their opinions.

Furthermore, the initial court order issued by Moraes to X lacked the specificity required by law. §1º states that any judicial order aimed at censoring content “must contain, under penalty of nullity, a clear and specific identification of the content alleged to be infringing, allowing for the unequivocal location of the material.” When X later leaked the contents of these court orders, it was revealed that they did not provide “clear and specific identification of any content” but merely identified accounts instead. For this reason, these court orders should be considered unsupported by Brazilian Internet law. But because there’s no judicial authority above the Supreme Court, there is no one to appeal to.

As X challenged these court orders, Moraes subpoenaed the company’s legal representative in Brazil to testify under the inquest and threatened him with imprisonment for noncompliance. As legal scholars have noted, the Brazilian Code of Civil Procedures prohibits criminal charges for disobedience if a lighter sanction is available. In this case, a fine would be a more appropriate response.

In response to the threats, X closed its office in Brazil, leaving the company without the legally required local representation. Moraes then held X liable for an article of the Civil Code (Law 10.406/2022, Article 1.134) that mandates foreign companies to appoint a legal representative to operate in Brazil. Moraes’ use of the Civil Code to demand that foreign companies appoint a local representative is also not entirely grounded: legal scholars argue that not every company is required to have a representative. Moreover, as other international platforms were also operating in Brazil without a local representative, this raises questions about the selective enforcement of the law. Ironically, after X was banned, the Supreme Court created an account in Bluesky, an alternative to X, which did not have a local representative.

The ban against X was issued based on the assertion that the company failed to have a representative in the country. However, Moraes’ decision extended even further. It imposed a daily fine of R$50,000 (approximately U.S. $9,000), which is higher than the average yearly income in Brazil, and potential civil and criminal penalties for any Brazilian who attempted to access X via a virtual private network (VPN). The ruling also mandated that Google and Apple remove any VPN applications from their app stores in Brazil.

Furthermore, in an effort to push Elon Musk to comply with previous court rulings, Moraes froze X’s and Starlink’s bank accounts, again without due process. He later seized the account balances to effectively force the companies to pay U.S. $3 million in fines that had been imposed so far.

The Brazilian Bar Association filed an Allegation of Noncompliance with a Fundamental Precept (ADPF), a formal complaint concerning the disregard for due process at the Supreme Court. Additionally, it issued an opinion opposing the imposition of a $9,000 fine on anyone who accessed X through a VPN, arguing that penalty fees “should never be imposed in advance and summarily”  and “without due process of law.” 

The origins of judicial overreach in Brazil

In March 2019, the Supreme Court initiated the “Fake News Inquest” to investigate an article published in the political blog O Antagonista, run by Crusoé magazine. In the article, Diogo Castor de Mattos, one of the main federal prosecutors in the nationwide corruption investigation now known as Operation Car Wash, accused the Brazilian Supreme Court of undermining parts of its federal investigation. Mattos was outraged by the attempts to move the case to the electoral courts, which lack the resources to handle such cases properly. Mattos called it a “blow against the operation quietly prepared by the Brazilian Supreme Court.”

This prompted then-President of the Court, Dias Toffoli, to open an investigation into “fake news and threats” directed at the court and its members. “The Fake News Inquest” had commenced, and it would soon become the center of controversies, as noted in an exceptionally accurate piece by The New York Times titled “To Defend Democracy, Is Brazil’s Top Court Going Too Far?”

In April 2019, Crusoé magazine published another piece reporting on leaked court documents related to Operation Car Wash. Some of these documents revealed that one of the main parties under investigation testified that a recurring alias during the inquiry referred to Judge Dias Toffoli.

Crusoé magazine had reported on publicly available information from the testimony. Yet, Toffoli then asked Moraes, who was then a judge (and is now the president of the court), to investigate the case under the authority of the Fake News Inquest. Three days later, Moraes ordered the censorship of Crusoé’s article covering the testimony.

After facing severe backlash, following condemnation from both the Brazilian Bar Association and other Supreme Court justices, Moraes eventually retreated and allowed Crusoé to publish the piece, but a precedent had been set. The court now had the power to issue orders at will to safeguard its reputation and integrity.

On the day after the conclusion of Crusoé‘s case, Attorney General Raquel Dodge requested that the inquest be closed and archived. Moraes and the Supreme Court blatantly ignored Dodge and continued investigations.

In August 2019, the attorney general was joined by the National Association of Federal Prosecutors in an attempt to declare the inquest unconstitutional. The attorney general and the association argued that: (one) The Supreme Court started the inquest ex officio; (two) Toffoli specifically nominated Moraes rather than selecting a judge via lottery; (three) the inquest violated the required separation between accusation and judging; (four) the investigations were confidential, such that the alleged suspects could not defend themselves, nor were they are aware of the alleged crimes for which they were accused; and (five) the inquest did not specify a timeframe for investigation, nor name any suspects.

The last complaint was the most critical fact to the attorney general, as it represented an open-ended and limitless blank check for the Supreme Court to investigate anyone, anywhere, anytime. However, the Supreme Court itself had the final say on whether their investigation was lawful and constitutional. Only one justice dissented in a 10 to 1 vote.

Judicial rule-bending

The investigative powers used in the inquest are based on Article 43 of the Supreme Court’s internal bylaws, which state that “following a criminal offense occurring within the premises or dependencies of the [Supreme] Court, the president shall initiate an investigation if it involves an authority or person under the Court’s jurisdiction or delegate this responsibility to another justice” (emphasis added).

It is unclear whether the powers awarded by the Supreme Court’s internal bylaws can be used to investigate whistleblowing by a federal prosecutor, not to mention Crusoé‘s piece. Neither are criminal matters, and none occurred “within the premises” of the Court. Toffoli argued that although the alleged crimes have not been committed inside the court, the alleged victims—the justices themselves—”are the court.” Moreover, in cases where the president “delegate[s] this responsibility to another justice, “this is usually done by a lottery—but Toffoli specifically chose Moraes for this one.

UOL, a national news outlet in Brazil, attempted to audit the algorithm that randomly assigns cases to each justice. While access to the system was initially granted, on the day journalists and technicians were set to test it, access was suddenly suspended without further explanation.

Suspicions of political motivations

Under the Fake News Inquest, Moraes ordered the banning of several social media accounts. It remains unclear who was banned and for what reasons, but many have noted that nearly all the banned accounts appear to belong to right-leaning politicians and media personalities associated with former president Jair Bolsonaro. Since the files are classified, all that is known comes from what X leaked to the public.

For instance, in the files leaked by X, it was revealed that Moraes ordered the banning of social media accounts linked to a lawful protest in New York City in November 2022 against him and the Supreme Court. One of those accounts belonged to David Sacer, a well-known gospel singer with millions of followers across his social media platforms, who shared a post announcing that Moraes would be giving a lecture in New York.

Folha de São de Paulo, the largest newspaper in Brazil, leaked informal communications between Moraes, his office staff, and the Supreme Electoral Court’s Special Advisory for Combating Disinformation (AEED in the Portuguese acronym). The leaked documents show that Moraes and his staff were ordering the AEED to manufacture “evidence” against the New York protestors to be used in the Fake News Inquest. Moreover, many of Moraes’ targeted requests for investigation were reported as if they arose from “anonymous tips.” As Folha alerted, this practice was common:

In several cases, the targets were chosen by [Moraes] or an assistant. The reports were adjusted when they were not satisfied and, in some episodes, were tailor-made to support predetermined action. When they couldn’t find clues against a target, one assistant guided the other: “Use your creativity… lol”

As mentioned earlier, one of the most notable revelations made by X when releasing the full content of the secretive orders received from Moraes was their vagueness. The orders did not point to any specific posts made by these accounts, nor did they specify which laws had been broken; they simply demanded that entire accounts be suspended indefinitely and warned the platforms not to disclose that the orders were originating from Moraes.

Many legal scholars in Brazil believe Moraes’s actions are unfounded. It is frequently noted that the decision lacks a basis in jurisprudence and conflicts with established legal principles. Furthermore, the orders were not aimed at removing content but at accounts—Brazilian law does not allow this, as it would prevent individuals from expressing their opinions beforehand and indefinitely.

According to X, Moraes has issued secret orders to take down over 200 profiles. The targets of these decisions range from Nikolas Ferreira, the most-voted federal congressman in Brazilian history and a member of the former President Jair Bolsonaro’s party, to other elected senators and completely anonymous individuals.

Another target was Luciano Hang, a businessman and campaign donor of former president Jair Bolsonaro, who has had his accounts with millions of followers suspended for more than two years without trial, formal indictment, or any opportunity to present a defense. The decision, as is often the case, did not identify any criminal content to justify the suspension. On September 3, 2024, Moraes suddenly decided to reactivate Hang’s accounts, warning him not to “reiterate the same illicit practices.” To this day, it is unknown what “illicit practices” Hung was accused of committing.

It is unknown how many people have been investigated by the Brazilian Supreme Court so far because most of its contents are classified. It’s also worth noting that after five years of the “Fake News” inquest, only one former congressman, Daniel Silveira, has been indicted, charged, and allowed to present a defense. Lawyers for those under investigation have repeatedly complained that they have no access to the “evidence” cited against their clients, and they are essentially in the dark while trying to defend them.

The Brazilian Constitution explicitly prohibits censorship, especially for political reasons. Article 220 states, “[T]he manifestation of thought, creation, expression, and information, through any form, process, or vehicle, shall not be subject to any restriction, as provided by this Constitution. … All and any censorship of a political, ideological, or artistic nature is prohibited.”

More censorship of social media companies: Rumble and Telegram

In 2023, the Brazilian Congress was preparing to vote on Bill 2630, which would establish state control mechanisms over social media and impose a new tax on “Big Tech” companies to fund traditional media outlets. Legal scholars strongly criticized the bill for introducing a “duty of care,” making social media platforms liable if they failed to address what was defined as “fake news.”

Moraes used the Fake News Inquest to summon executives from Google, Meta, and Spotify to testify at the Federal Police headquarters. Acting independently, Moraes asserted––without any supporting evidence––that Google was manipulating search results to favor web pages that labeled the social media regulation (Bill 2630) as censorship. Moraes also ruled that the social media platforms could not issue statements criticizing the issue.

Further, Moraes ordered Google to remove a post titled “How PL [Bill] 2630 Could Worsen Your Internet” from its own blog. Moraes also required Telegram to remove a message it had sent to users in Brazil, urging them to contact their elected representatives in opposition to the bill. Moraes then forced Telegram to send another message written by him instead, compelling the platform to admit guilt and declare that the original message was fake news. Despite this (or perhaps because of it), public pressure led Congress to retract the bill, which is now archived.

Legal orders against social media companies have become ubiquitous. Telegram has been temporarily banned twice, once in February 2022 and then in April 2023. Rumble was permanently banned on February 20, 2025. In the 2023 Telegram case, a federal court ordered Telegram to share the contents of conversations from a neo-Nazi group, which is “technologically impossible” due to end-to-end encryption.

Rumble and Telegram were also banned in Brazil for failing to comply with Supreme Court orders. In these cases, the orders required the banning of only one account, political blogger Allan dos Santos. Although he has been residing in the United States since 2020, Moraes issued an arrest warrant against him, froze his bank accounts in Brazil, and requested his extradition from the U.S., but the prosecutor general opposed the measure.

Moraes’s attempts to reach outside of Brazilian jurisdiction are common. Not only did he subpoena Elon Musk through a tweet, but he also ordered X to ban the account of a Brazilian-born U.S. citizen residing in Miami, Florida, and then issued an illegal arrest warrant against her.

Incoming decisions to impose further deterioration of free speech in Brazil

The prospects for free speech and internet regulation in Brazil are increasingly troubling. As of December 4, 2024, the Brazilian Supreme Court has begun deliberating the constitutionality of Article 19 of the Brazilian Internet Law (Marco Civil da Internet Law 12.965), which could lead to its repeal.

Early opinions suggest the following implications of a repeal of Article 19:

1. Platforms must immediately remove content deemed “offensive to honor, image, or privacy” upon notification or face legal liability for failing to prove it wasn’t offensive in court. As Justice Luiz Fux stated, “Once notified, removal must be immediate, and the platform must seek judicial authorization to reinstate it.”
2. Because of that, users will be unable to defend themselves, and it would be the platform’s duty to legally appeal on their behalf to prevent a ban.
3. Platforms will be liable, even without notification, for issues like “fake accounts” and “risks to fundamental rights or electoral integrity.” Since these terms are vague, platforms may aggressively remove related content to avoid penalties, as Meta has stated.

These changes would substantially undermine due process, potentially forcing platforms to comply with arbitrary demands for censorship. Revoking Article 19 would expand the Brazilian Supreme Court’s power to control online discourse, censoring and discouraging free expression. If the court moves forward with these changes, it will mark a significant shift toward greater state control of online platforms, further eroding Brazil’s democratic institutions.

The court will likely view Article 19 as insufficient to enable its desired interventions. However, inefficiency does not make it unconstitutional. The Brazilian Congress should address the former. Nevertheless, after the failure of Bill 2630, the justices might use this ruling to implement similar measures, such as the “duty of care” for content publishers—that Congress and the Brazilian people previously rejected—through judicial means, effectively bypassing the legislative branch.

Additionally, this ruling could retroactively legitimize the court’s current modus operandi, including Moraes’s secretive orders. These orders are at odds with current Brazilian law, as the law does not allow the ban of full accounts, much less without due process or evidence of any wrongdoing, and the constitution explicitly forbids politically motivated censorship. While the court has left the trial on standby, it may resume at any time.

Brazil’s internet governance outlook

​In February 2025, the U.S. State Department’s Bureau of Western Hemisphere Affairs issued a statement expressing concern over actions taken by Brazilian Supreme Court Justice Alexandre de Moraes. The Bureau criticized the blocking of access to information and the imposition of fines on U.S.-based companies for refusing to censor individuals residing in the United States, stating that such measures are incompatible with democratic values, including freedom of expression. Brazil’s Ministry of Foreign Affairs released a statement rejecting any attempt to politicize judicial decisions, defending the Supreme Court’s actions as simply enforcing Brazilian laws.

Brazil accounts for a significant share of the social media market. It is home to 144 million social media users, representing 66% of its total population. The country is the world’s fifth-largest social media market. In terms of daily internet usage time, Brazil ranks second, with Brazilian users spending an average of 9 hours and 13 minutes per day online.

The current legal landscape, shaped by judicial overreach and discretionary authoritarian control, poses a significant threat not only to social platforms and their businesses in the country but also to the very foundations of democracy and free speech.

Brazilian democracy has indeed encountered some turbulence. Allegations of electoral fraud regarding the 2022 election, which removed Bolsonaro from power and brought Lula da Silva back, incited an attack on Congress by right-wing activists on Jan. 8, 2023, who called for a military coup to restore Bolsonaro to power (a copycat of the Jan. 6 attacks in the U.S.).

Moraes has referenced this and other instances of political extremist supporters of Bolsonaro to retroactively justify the continuation of the investigation and the court’s discretionary use of power. The court claims to protect Brazil’s democratic institutions, but its increasing control over political discourse and suppression of dissenting voices have the opposite effect. The actions of the Brazilian Supreme Court, particularly under Justice Moraes, have set a dangerous precedent for judicial censorship and overreach.

If the Supreme Court continues to expand its powers and erode legal protections for free expression, Brazil risks becoming a more authoritarian state where the judiciary, rather than the legislature, arbitrarily dictates the boundaries of free speech.

There’s no evidence the Brazilian Supreme Court intends to give up the extraordinary powers it has consolidated. Even if the Fake News Inquest concludes, the court now has a precedent to establish new unchecked “inquests” that could operate similarly, disregarding due process or the rule of law.

Social media platforms and internet providers in Brazil are increasingly seen as tools to enforce the court’s whims, with little regard for legal principles. The court often requires orders to be complied with within a few hours and imposes increasingly hefty fines, leaving little room for appeals. As the Supreme Court is the top judicial authority, there is no one else to appeal to.

This unpredictable and highly politicized environment creates an atmosphere of fear and compliance, where companies like X must choose between complying with court demands or abandoning service to the country altogether.

All internet and social media companies operating in Brazil are now at the mercy of Chief Justice Moraes and the Brazilian Supreme Court, with little to no protection under the rule of law. Given the court’s arbitrary and capricious rulings, these companies can expect to continue being used to fulfill the court’s personal and political agenda. A reversal of Article 19 of the Brazilian Internet Law (Marco Civil da Internet) would drastically exacerbate this threat.

What makes this especially alarming is not just the court’s conduct, but the international support it has received. Framing judicial censorship as “digital sovereignty” risks legitimizing a model of unchecked, discretionary power over online speech—one that could embolden courts and political actors, particularly in authoritarian regimes, to also impose arbitrary censorship on social media platforms under the guise of governance. The latter has already been tried in the US under the extensive lawfare aimed at censoring dissenting views in social media during the pandemic. The modus operandi was the same applied first in Brazil as early as 2019: covert pressure on social media platforms such as Facebook and Twitter.

The post How Brazil’s X ban signals growing control over online free speech appeared first on Reason Foundation.

Source: https://reason.org/commentary/how-brazils-x-ban-signals-growing-control-over-online-free-speech/