Brazil is regulating AI use in elections. Here’s why it’s been a challenge.

Rio de Janeiro | Oct. 03, 2026, 5:00 a.m. ET | 5 min read Dona Maria is a straight-talking, middle-aged Black woman, whose videos criticizing the policies of Brazilian President Luiz Inácio Lula da Silva amassed millions of views on social media earlier this year.“They line their pockets with our sweat!” she says in one clip about a supposed increase in electricity bills.But “Dona Maria” isn’t a Brazilian voter – or even human. She’s a hyperrealistic avatar, created by artificial intelligence. As countries around the globe are voting in increasingly AI-saturated elections, Brazil is on the front lines of trying to regulate the technology’s use in the electoral context. Its election authority, the Superior Electoral Court, passed broad rules this year, including requiring AI-generated campaign material to carry clear labels and prohibiting the use of AI to impersonate candidates.In the United States, at least 31 states have enacted local legislation with similar rules. But globally, election regulation regarding AI has “been perceived as a back burner issue” compared with other AI threats, says J.B. Branch, director of federal AI governance and technology policy at Public Citizen, a U.S. consumer advocacy organization. Facebook screengrabA Facebook reel of AI-generated avatar "Dona Maria" criticizes a planned increase to electricity bills, Sept. 24, 2026. Brazil’s polarized political landscape is rife with disinformation – a challenge that can be turbocharged by AI. From deepfakes of well-known journalists to satirical videos criticizing candidates and official campaign materials, how AI use is playing out this campaign season underscores the challenges it can present to even the most proactive governments. Avatars such as Dona Maria, sometimes dubbed “synthfakes,” are part of the constantly evolving AI ecosystem here.Not all AI use is negative, says Matheus Soares, head of content at Aláfia Lab, one of two civil society organizations running a monitoring project tracking the use of AI in the elections. But cases like Dona Maria are of particular concern and difficult to regulate. “These are very realistic characters commenting on politics,” he says. “They can have an impact on public debate, on citizens’ perceptions of a candidate, or a political topic.”Disinformation and AIOn Oct. 4, 159 million registered voters in Brazil will choose their next president, as well as senators, federal lawmakers, state governors, and state representatives. Leading the presidential race are Lula, as the current head of state is widely known, and Sen. Flávio Bolsonaro, a son of former President Jair Bolsonaro. Polls have the two candidates neck-and-neck. From India to Germany to Hungary, countries around the world have been grappling with artificial intelligence’s impact on electoral processes since AI tools went mainstream over the past couple of years. It’s a particularly relevant conversation in Brazil, a hyper-online country where disinformation campaigns dominated the past two presidential elections, in 2018 and 2022.This year, “generative AI is the big novelty,” says Diogo Rais, a professor of electoral law at the Mackenzie Presbyterian University in Sao Paulo. While AI’s uses can range from content production to research purposes, Brazil’s top electoral court is particularly concerned about its power to fuel fake news, he says.“Brazil has a very challenging history. In 2018, we were bulldozed by disinformation,” says Camila Tsuzuki, head of research at InternetLab, a nonprofit Brazilian research center. That year, fake news discrediting candidates on the left – associating them with pedophilia or plans to indoctrinate schoolchildren with “gender ideology,” for example – flooded messaging apps such as WhatsApp. Far-right populist Jair Bolsonaro ended up winning.Four years later, when Mr. Bolsonaro lost his reelection bid to Lula, disinformation about the integrity of the electoral process culminated in the Jan. 8, 2023, attacks on government institutions in the capital, Brasília. Many of the Bolsonaro supporters who stormed the capital said they believed the election had been rigged, repeating unfounded claims of electoral fraud peddled by Mr. Bolsonaro himself. The former president is now serving a 27-year sentence under house arrest for his role in plotting the coup attempt.“Borderline impossible”On the streets of Rio de Janeiro, even social-media-literate voters believe they run the risk of being deceived by AI.“If you don’t pay attention to the detail, you’ll believe what you see,” says Ana Luiza Lelis, on her way home after a day’s work passing out leaflets for a gubernatorial candidate’s campaign. In the run-up to this year’s campaign, the AI in Elections Observatory recorded 29 cases of AI avatars like Dona Maria, though no other synthfakes obtained her level of notoriety. An April analysis by BBC News Brasil found that Dona Maria’s online content was getting similar engagement levels to those of some national politicians. Constance MalleretCandidate Rafael Satiê speaks at campaign event, with an AI-generated drawing of Flávio Bolsonaro in the background, in Rio de Janeiro, Sept. 19, 2026. Mr. Satiê says he avoids using generative AI for his campaign materials. Lula’s coalition filed a legal complaint requesting that the avatar’s profiles be taken down, on the grounds that the anonymous accounts peddled false or distorted information that could mislead voters.“Physical characteristics create a rapport with the person receiving the information,” says Ms. Tsuzuki. “And the greater the rapport, the greater the likelihood that someone will be open to the information or the persuasion of that message.”Rafael Satiê, a Rio de Janeiro city councilor running for federal deputy on a pro-Bolsonaro platform, says his communications team generally steers clear of generative AI. “I prefer not to use [AI], to avoid running into any issues,” says Mr. Satiê, before heading into a campaign event flanked by real-life social media influencers.Brazil has some of the strictest rules around the use of AI in elections because of the electoral court’s power to pass regulations and resolve legal disputes, meaning it doesn’t depend on slow legislative processes, says Dr. Rais, the law professor. In the United States, bipartisan legislation to protect elections from deceptive AI has failed to make it through Congress, while in Brazil, a bill for AI regulation beyond elections remains stuck in the lower house.But even these election rules are hard to enforce, not least because of the endless nature of online content, and the legal gray area cases like Dona Maria sometimes fall into. The character’s main TikTok and Instagram accounts were taken down by the social media platforms in May, but secondary accounts have since popped up, and some content remains available online. “Trying to regulate individual users is borderline impossible,” says Mr. Branch.“The big question”Brazil’s electoral rules impose a blackout on posting and resharing AI political content in the 72 hours before the vote and the 24 hours that follow it.But observers worry about another, less visible phenomenon: the influence of chatbots such as ChatGPT, Anthropic’s Claude, or Google’s Gemini. A growing body of scientific research is delving into AI’s powers of persuasion, and several experiments in the United Kingdom and the United States have found that chatbots can shift people’s political views.On this, too, Brazil has electoral laws that ban AI tools from ranking candidates or emitting opinions. But that doesn’t stop some chatbots in Brazil from doing those things, according to research carried out by InternetLab at the end of August. Deepen your worldviewwith Monitor Highlights.Politics with respectGet political stories with respectful analysis.There‘s a world of new ideas in everyBooks newsletter.There‘s more to life, enrich yours withCulture & Learning weekly.Follow humanity‘s discoveries withScience & Nature stories in your inbox.Gain a spiritual perspectivefrom the stories in your inbox.Want to understand the deeper impact of critical events? Learn the Monitor‘s insight.Already a subscriber? Log in to hide ads. How much these processes and platforms are interfering with voters’ decision-making is going to be “the big question of these elections,” says Ms. Tsuzuki.It’s “going to be something of a stress test.” ALREADY A SUBSCRIBER? LoginReal news can be honest, hopeful, credible, constructive.The Christian Science Monitor was founded in 1908 to lift the standard of journalism and uplift humanity. We aim to “speak the truth in love.” Our goal is not to tell you what to think, but to give you the essential knowledge and understanding to come to your own intelligent conclusions. Join us in this mission by subscribing.