As Brazil prepares for a closely contested presidential election on Sunday, an unprecedented surge of artificial intelligence-generated videos has flooded social media platforms, intensifying concerns over disinformation in the country’s highly polarized political environment. These AI-created videos range from humorous to misleading, reflecting a rapidly evolving challenge for Brazil’s election authorities and setting a significant precedent for the global electoral use of such technology.
Brazil, the largest democracy in Latin America with a population of 213 million, has long grappled with widespread online disinformation. However, the growing sophistication and volume of AI-generated content are testing the robustness of the nation's stringent electoral laws, which include prohibitions on “deepfakes” and require clear labelling of AI-produced media during campaigns. The Superior Electoral Court, responsible for overseeing the polls, has enacted strict regulatory measures aimed at preserving the integrity of the vote, although these rules are largely untested until the current election cycle.
Chief Justice Kássio Nunes Marques of Brazil’s electoral court emphasized the challenge of distinguishing authentic political communication from digitally altered content, questioning whether voters can trust AI-enhanced messages. Social media companies such as Meta, Google, and OpenAI have collaborated with authorities by submitting plans to manage AI-generated political content and by implementing measures like automated watermarking and biometric detection to identify manipulated videos. Despite these efforts, independent observers report violations, including unlabelled AI content from political campaigns and platforms continuing to recommend chatbots connected to candidates in contravention of court orders.
The rapid dissemination of AI-generated disinformation outpaces the relatively slow and bureaucratic process of content removal mandated by the electoral court, complicating enforcement efforts. Experts warn that social media algorithms, which personalize content and prioritize engagement, can amplify misleading AI videos, making disinformation more pervasive and harder to control. At the same time, criticism has emerged regarding the impartiality and moderation effectiveness of major technology firms, with accusations from President Luiz Inácio Lula da Silva’s camp and watchdog groups that some platforms disproportionately favor his opponent, Flávio Bolsonaro.
The political contest pits Lula, the incumbent leftist president, against his right-wing rival Flávio Bolsonaro, son of former President Jair Bolsonaro, who remains a controversial figure after his recent imprisonment and continued influence over his political base. AI videos have frequently featured fabricated scenarios, such as false polls, staged crowds, and dramatized depictions of both candidates to sway public opinion. For instance, manipulated footage portrays Lula as corrupt and fleeing police scrutiny tied to a Supreme Court corruption scandal, while Bolsonaro is sometimes cast as a superhero combating crime and restoring the economy.
Brazilian voters’ familiarity with memes and high social media engagement, combined with relatively high trust in AI technologies compared to other countries, has contributed to the traction of these videos. Analysts note that AI has not created new types of political falsehoods but has accelerated their production and dissemination, lowering barriers for actors seeking to influence the electorate.
Moreover, the impact of AI extends beyond video manipulation. Popular chatbots providing personalized political advice are emerging as influential tools capable of shaping voter behavior, raising questions about transparency and accountability. Luiz Alexandre Reali Costa, director of the Brazilian Observatory for Artificial Intelligence, observed that while AI’s challenges are not fundamentally novel, they amplify existing vulnerabilities in Brazil’s electoral ecosystem.
As the nation goes to the polls, Brazil’s experience underscores the growing global struggle to address AI-driven disinformation in democratic processes and the urgent need for adaptive regulatory and technological responses.