Subscriptions
Read our Monthly Magazine
And support our mission to provide fearless stories about and outside the media system
Far-right populists worldwide are using AI to bypass traditional media gatekeepers and spread disinformation at scale. Nowhere is that more true than Brazil, which has been rocked by political disinformation in recent years and is expected to face even more daunting challenges with the rise of generative AI. These concerns are now reaching an inflection point just days ahead of Brazil’s general election on October 4 — a contest that sits at the intersection of the race to influence democracy with AI and the race to control the technology itself.
Thirteen candidates are running for the presidency, but the race is currently dominated by incumbent President Luiz Inácio Lula da Silva of the Worker’s Party (PT) and Senator Flávio Bolsonaro of the right-wing Liberal Party (PL). Lula won by a narrow margin in 2022, just edging out Jair Bolsonaro (Flávio’s father), who is now serving time for his role in plotting a coup in an attempt to overturn the 2022 election results.
The upcoming election mirrors other recent elections in the region, with high levels of distrust in democratic institutions and the possibility of another right-wing candidate taking office and aligning closely with the presidency of Donald Trump.
Brazil is also a strategically important location, particularly for AI companies, as it is becoming a priority market for data-center investment and is a crucial source of critical minerals that are needed to power and build AI. The United States and European Union, as well as private companies, are now looking to Brazil as they try to reduce supply-chain dependence on China, and Bolsonaro has indicated his support for such a realignment.
Because of this, access to Brazil’s Government is both politically and commercially important, making the upcoming election a prime target for interference by foreign governments and tech companies alike.
Information Warfare in Brazil
For the past several election cycles, Brazil has served as a testing ground for information warfare tactics in the digital age. The digital environment there has been described as “a fertile ground for innovative strategies to spread polarizing content with the intent of manipulating public debate.”
Several factors converge to make Brazil uniquely vulnerable to political disinformation and malign influence, including a heavy reliance on hyperconnected chat applications, intense political polarization, and a powerful influencer-driven media market.
Brazil was early to recognize the potential dangers of AI as a tool of election influence and interference. Ahead of its 2026 election, the Superior Electoral Court, known as the TSE, prohibited generative AI systems from providing certain election-related responses, including those that recommend or prioritise candidates, parties, and campaigns. The rules also prohibit these systems from expressing an electoral preference or recommending a vote, and impose restrictions on AI-generated content.
In theory, those rules are supposed to help prevent AI systems from interfering in elections via misinformation, disinformation, or content plagued by bias, partisanship, or other forms of hidden influence. In practice, however, that’s not how things are playing out.
Research on these systems has found that most of the major AI models ranked or ordered candidates in nearly every instance during testing, and official sources were not reliably used when doing so. Additionally, responses were often inconsistent and varied based on the wording of the input, meaning that voters who turn to these systems for information could receive differing responses with variable source quality, yet no one would know because no one has access to view all of these responses, except for the companies themselves.
In addition to AI chatbots violating Brazil’s election rules, Meta’s ad system was found to be approving advertisements containing content that violates both Meta’s own community standards and national election laws. Among the prohibited content that was approved were ads calling for a repeat of the January 8th attempted coup in Brazil and another inciting violence against a lawmaker. “Other approved ads targeted lawmakers with hate speech, falsely celebrated Trump’s official endorsement of Flávio Bolsonaro, and pushed the conspiracy theory that Jair Bolsonaro is a political prisoner,” according to corporate accountability watchdog Eko.
Revealed: Reform UK Crypto Partner’s Ties to Pro-Trump Big Tech and Kemi Badenoch
Nigel Farage’s cryptocurrency partner is tied to Trump-supporting tech firms and senior Conservative figures, reports Nafeez Ahmed
Big Tech Sets its Sights on Brazil
As a result of these risks and the Government’s fractured efforts to contain them, Brazil has found itself in the grips of a major battle between state regulators and global tech giants.
Brazil’s Superior Electoral Court and Supreme Court have imposed some of the most aggressive anti-disinformation measures globally, implementing strict frameworks that ban deepfakes and enforce platform liability. However, this top-down approach to regulation has turned the digital environment into a proxy war over sovereignty and free speech, which is exemplified by high-profile clashes between Brazilian authorities and tech platforms. This conflict is frequently weaponised by political actors to fuel anti-establishment narratives, compounding public distrust in the information ecosystem.
In 2024, after extensive corporate lobbying, the Senate in Brazil removed previously proposed safeguards against high-risk AI models, reportedly in response to pushback from tech giants. A year prior, Google and Telegram had come under investigation by Brazilian authorities for allegedly interfering with Brazil’s Fake News Bill, which sought to hold platforms accountable for misinformation. Much like they’ve embraced the Trump administration in the US, tech companies have forged an alliance with the far-right in Brazil in an effort to avoid regulation.
As part of this battle in 2024, Elon Musk challenged Brazil’s Supreme Court over its attempts to block X accounts associated with disinformation networks and even encouraged Brazilians to defy the courts by using virtual private networks (VPNs) to circumvent the blocks. In response to this, the court included Musk as a target in its investigation into the spread of disinformation and opened up a separate inquiry of his conduct, which the court alleged amounted to obstruction. In August 2024, the court blocked X entirely, though that restriction was removed two months later after the platform complied with a court order.
Musk’s confrontation with Brazil’s Supreme Court has supplied the Bolsonaro movement with a powerful international champion. At a 2024 rally, Jair Bolsonaro praised Musk while supporters displayed signs thanking him. The former president went on to praise Musk, describing him as “the man who really preserves true freedom for all of us” — a statement that tapped into the culture war over free speech and enlisted far-right Brazilians as fellow warriors.
In January 2025, Meta CEO Mark Zuckerberg joined in the attacks, insinuating that “secret courts” in Latin America are carrying out a mass censorship campaign, which was widely interpreted as a reference to Brazil’s Supreme Court. Zuckerberg has also pledged “to work with President Trump to push back on governments around the world that are going after American companies”, referring to efforts by state regulators to reign in Big Tech.
Personalising Propaganda
Technology companies have influenced elections for years, both directly and indirectly. The public gets information through search engines, then shares it on social media, where different segments of the population are then exposed to targeted political advertising based on profiles developed using their online data. With disinformation and misinformation playing an increasingly important role in election cycles, tech companies are also at the forefront of the battle over content moderation and restrictions on speech.
Brazil has been at the centre of much of this, including during the previous two election cycles, when false claims about elections, voting, and related issues circulated across social media and messaging platforms, particularly WhatsApp.
Many of the existing concerns about WhatsApp are still relevant today, even though those concerns have now grown to include AI chatbots as primary sources of bias and false information. Just like WhatsApp, AI chatbots deliver content privately, and are largely inaccessible to the public or to any efforts to audit the information they produce. On WhatsApp, political campaigns in Brazil were found to be operating vast numbers of groups that were made to appear distinct in order to conceal the fact that a few administrators were controlling most or all of them. AI chatbots operate from a similar black box, with most users in the dark about where the chatbot is actually getting its information or how it chooses what information to prioritise or deprioritise.
Just like WhatsApp in 2018 and 2022, AI chatbots represent a fundamental shift in how Brazilian voters are getting information that may influence who they vote for. Increasingly, that information is being delivered via private and highly personalized channels that can’t be viewed by auditors, election observers, or fact-checkers. No one knows exactly how much mis- and disinformation voters are being exposed to via chatbots in the run-up to the election because no one even knows what information people are seeing when they access it via an AI-mediated platform.
This is particularly problematic given that digital platforms are known to disproportionately amplify right-wing propaganda and have repeatedly allowed partisan bias to influence decisions about the content that users can access.
On September 23—just 11 days before the election—Meta temporarily prevented Lula’s campaign page from running political advertisements and removed an Instagram account that had published reporting about Flávio Bolsonaro’s alleged connections to financial scandals. Meta later reversed course, but its explanation for the restriction was unclear and the damage was already done. Three years earlier, Google used its Brazilian search homepage to direct users toward opposition to proposed platform regulation and encourage them to lobby lawmakers. Google defended the decision and denied manipulating search results.
ENJOYING THIS ARTICLE? HELP US TO PRODUCE MORE
Receive the monthly Byline Times newspaper and help to support fearless, independent journalism that breaks stories, shapes the agenda and holds power to account.
We’re not funded by a billionaire oligarch or an offshore hedge-fund. We rely on our readers to fund our journalism. If you like what we do, please subscribe.
Bolsonaro’s ‘Digital Armies’
While consequential decision-making within tech companies remains opaque, one thing that is clear is that AI-generated election disinformation has been proliferating in the days and weeks ahead of the election. In just the first 11 days of the campaign, the TSE received 19 cases involving the use of AI in political advertising and campaigning.
In July 2026, Flávio Bolsonaro rolled out an AI-generated campaign video featuring deepfake footage of his father, who is currently under house arrest, making it appear as if he was out on the campaign trail with his son. The elder Bolsonaro is barred from making any type of public statements or using social media, but his son appeared to get around those restrictions by using AI-generated depictions of his father to create the appearance of an endorsement. The former president is still a popular figure in Brazil, so having his endorsement is widely seen as an important mark for his son’s campaign.
Left-wing parties challenged the video, arguing that it illegally evaded restrictions imposed on Jair Bolsonaro. Brazil’s Superior Electoral Court ultimately ruled, by four votes to three, that the video did not count as public electoral advertising because it appeared before Flávio’s candidacy was formally registered. The court then issued a broader definition of prohibited deepfakes for the remainder of the campaign season. Flávio has additionally produced at least one AI-generated campaign song and has shared several AI campaign videos.
Supporters of Bolsonaro have also reportedly organised massive digital “armies,” including a WhatsApp group where members receive daily “visual ammunition” to post on social media in an effort to boost Bolsonaro’s candidacy. The accompanying website reportedly “contains ready-made posts, allowing users to filter content by their preferred social media platform or category.” One of the largest categories of content on the website focuses on “criticism of the PT” (the Workers’ Party, represented by Lula). This mirrors previous astroturfed digital campaigns led by members of the Bolsonaro movement — the very campaigns that incited an attempted coup several years ago.
Whether or not this year’s digital armies try to fuel another insurrection remains to be seen, but what we do know for sure is that tech billionaires have a massive stake in the result of the election — and powerful tools to shape what voters see. As was recently demonstrated, AI can put a sidelined leader back on the campaign trail, and platforms can amplify a political attack or interrupt a campaign’s advertising. Whoever wins the upcoming election will help write the rules that tech companies must follow. The danger is that the people trying to win power and the companies helping them reach voters seem to find accountability equally inconvenient.