Brazil’s elections are among the most significant political events of 2026. As the electoral calendar advances, potential candidates begin positioning themselves early, seeking visibility and public support.
Meanwhile, the political environment has become increasingly vulnerable to misinformation. Distorted or false content is often used to gain electoral advantage or undermine opponents, reinforcing a scenario in which truth competes with “strategic narratives”.
In this context, artificial intelligence has emerged as a growing concern. The technology has made it easier to produce highly realistic images, videos and audio, lowering the barrier for the creation and spread of misleading content, often amplified on social media, where it can quickly reach and influence voters.
Deepfakes and the growing risk to electoral integrity
With the popularisation of artificial intelligence tools, such as voice cloning tools, face-swapping apps, image generators, and even AI for “fake news,” the production of manipulated content has become both accessible and scalable. So-called deepfakes, synthetic audio, images and videos, can distort reality in ways that are increasingly difficult to detect.
In a country approaching an election year, the implications are significant. The spread of false information poses a direct risk to Brazil’s democratic process, shaping public opinion and potentially affecting electoral outcomes.
According to DataSenado, 81% of Brazilians believe fake news can influence election results. This perception is linked to the speed at which manipulated content spreads across social media, often appealing to emotions and making it difficult to distinguish between reliable information and disinformation.
Part of the challenge lies in how this content circulates. Misleading material is often shared without full awareness of its consequences. What may begin as a joke in small groups can quickly gain traction, reaching large audiences with little verification. Surveys indicate that nearly 90% of Brazilians have believed in some form of false information, underscoring the population’s vulnerability.
Digital platforms further intensify this dynamic. Content that triggers strong emotional reactions — such as outrage, fear or surprise — tends to be prioritised by algorithms. Deepfakes, by their very nature, fit this pattern, allowing them to spread rapidly and widely.
Beyond organic circulation, there is also strategic use. Manipulated content can be deployed to reinforce political narratives, construct favourable public images or damage opponents. In this sense, disinformation becomes not just a byproduct of the digital environment, but an instrument of political dispute.
The consequences extend beyond individual cases. As misleading content becomes more prevalent, distinguishing fact from fabrication grows increasingly difficult. This overload weakens the quality of public debate and creates space for competing versions of reality, some of them more appealing than accurate.
There is also a broader impact on trust. In an environment saturated with manipulated content, even legitimate information may be questioned. This erosion of confidence allows public figures to dismiss authentic material as false, complicating accountability and undermining transparency.
Social inequalities can further deepen the problem. Groups with limited access to reliable information or media literacy are more exposed to manipulation, which may affect their participation in the democratic process.
Repetition also plays a key role. Information, even when false, can begin to feel true simply because it is frequently encountered, a phenomenon known as the “illusory truth effect”.
Taken together, these factors point to a shift in the nature of electoral disputes. Campaigns are no longer defined solely by proposals and debates, but also by the circulation of digital narratives — increasingly shaped by artificial intelligence.
Between rules and manipulation
Most AI-generated political content circulating in Brazil lacks transparency. Data from the AI in Elections Observatory indicates that two out of three synthetic materials shared on social media do not disclose their artificial origin.
A separate survey analysing 137 cases between December 2025 and February 2026 found that only 27% included some form of identification, such as watermarks or captions. The majority, including content posted by politicians, was published without any indication of AI use.
The impact has reached prominent political figures. President Luiz Inácio Lula da Silva was the most frequent target, appearing in 46 manipulated pieces of content. Cases involving Jair Bolsonaro, Nikolas Ferreira and Flávio Bolsonaro have also been identified, alongside materials targeting Supreme Federal Court justices such as Alexandre de Moraes.
In one instance, Flávio Bolsonaro shared a video combining real news footage with AI-generated images portraying Lula alongside armed criminals. Although the content included a watermark, the case illustrates how such material can be used to construct negative associations and influence voter perception.
Electoral authorities have begun to respond. In the first half of 2024 alone, during Brazil’s municipal election cycle, 43 cases involving the use of artificial intelligence in pre-campaign activities were judged. In nearly a third of them, the use was deemed irregular.
In Tutóia, in the state of Maranhão, a manipulated video used the image of journalist Renata Lo Prete alongside fabricated audio to simulate a negative news report about a local politician. Other incidents include the circulation of offensive and sexually explicit content targeting public figures, highlighting the technology’s potential for personal attacks.
In response, Brazil’s Superior Electoral Court (TSE) has introduced stricter rules. AI-generated content must now be clearly identified, with penalties including fines and removal. The use of deepfakes that simulate candidates’ actions or statements with the intent to mislead voters has been explicitly prohibited.
The court has also partnered with social media platforms to accelerate the removal of misleading content and expanded reporting mechanisms. Even so, enforcement remains a challenge.
Detecting manipulated material is not always straightforward, particularly as techniques become more sophisticated. In many cases, judicial decisions are issued only after the content has already reached a wide audience.
The speed of social media compounds the problem. Misleading posts can spread within minutes, while institutional responses often take longer to produce tangible effects. The growing number of cases suggests that artificial intelligence is no longer just a technological innovation in Brazil’s electoral landscape, but an emerging risk factor for the country’s democratic process.
______________________________
The article above was edited by Isabela Raucci.
Liked this type of content? Check Her Campus Cásper Líbero for more!