Brazil’s 2026 elections will take place in a digital environment marked by the growing presence of artificial intelligence tools capable of creating, modifying and distributing synthetic images, videos, audio and text. If, in 2024, the political use of these technologies still appeared in a more experimental form, the scenario taking shape for 2026 is one of greater incorporation of AI into political communication and into disputes over information.
The monitoring carried out by the Observatório IA nas Eleições (AI in Elections Observatory), an initiative of Data Privacy Brasil and Aláfia Lab, helps gauge the scale of this transformation. Between December 2025 and February 2026, it identified 137 cases of AI use related to political matters or to the recreation of political figures on social media, an average of 1.5 cases per day and a 50% increase over the daily average recorded the previous year. At least 34% of the posts were made by political actors, including politicians and parties.
Beyond the volume, what stands out is the way these technologies have been used. Synthetic content was used both to produce satire and criticism and to spread disinformation, attack reputations, create false narratives and produce content in support of politicians. The survey identified, for example, videos and audio simulating statements by public figures, fake election polls and manipulated images of politicians. In some cases, AI was used to create fictitious identities and simulate political support.
This scenario poses an additional challenge to the integrity of the electoral process: it will not always be evident to voters when content has been artificially produced or manipulated. As artificial intelligence tools make creating synthetic images, videos, audio and text more accessible and realistic, it also becomes harder to distinguish authentic content from content that has been fabricated or altered.
In an electoral context, this difficulty can have significant consequences, especially when synthetic content is used to portray candidates, simulate events, attribute statements that were never made or build narratives capable of influencing public perception. The problem, therefore, lies not only in the possibility of producing misleading content, but in the growing difficulty of recognizing it as such. In this context, transparency about the use of artificial intelligence becomes an important element in enabling voters to critically assess the information they receive and make more informed decisions.
The issue becomes even more relevant when false content takes on the appearance of real events. The Observatory identified, for example, AI-generated videos that simulated news reports and presented false information about election polls, as well as content that fabricated events that never took place involving political figures and members of the Federal Supreme Court.
At the same time, artificial intelligence is also taking on a more direct role in the experience of political information itself. Chatbots are already being used as tools to search for information about candidates, proposals and topics of public interest. In tests carried out by the Observatory with five chatbots available in Brazil, all of the tools identified and presented profiles of presidential pre-candidates. Four of them also ranked proposals in different areas, and four recommended candidates according to specific voter profiles.
It is in this context that the new rules of the Superior Electoral Court gain importance. For the 2026 elections, Resolution No. 23,755/2026, which updated the regulation of electoral advertising, established specific rules for providers offering artificial intelligence systems or equivalent technology. During the electoral advertising period, these providers may not rank, recommend, suggest or prioritize candidacies, parties or campaigns. The rule also prohibits automated responses that express opinions, indicate electoral preference, recommend votes or promote political or electoral favoritism or disfavor.
The regulation also sets out obligations related to the labeling of synthetic content. According to the Observatory, Article 9-B of Resolution No. 23,755/2026 requires that multimedia content produced or manipulated by AI and used in political advertising be identified in an explicit, prominent and accessible manner, indicating that the material was fabricated or manipulated and the technology used.
The existence of the rule, however, does not settle the debate. The Observatory’s monitoring shows precisely the gap that can exist between the legal provision and the actual circulation of content on platforms. The presence of unlabeled synthetic posts, including among content published by political actors, raises questions about transparency, accountability and enforcement capacity during the electoral period.
There is also a dimension related to political and gender-based violence. Among the cases identified by the Observatory was content that used AI to manipulate images of women in politics, produce sexual deepfakes or create material intended to attack their reputations. The sophistication of these tools can amplify existing forms of political violence, especially when manipulated content is produced to look authentic and may continue circulating even after being removed.
The debate on AI in elections is therefore not limited to the question of what is true or false. It also involves who produces a given piece of content, how it is identified, what distribution mechanisms are used, what criteria guide automated systems and how synthetic content can influence perceptions, reputations and political choices.
It is against this backdrop that the Observatório IA nas Eleições holds its workshop at the Data Privacy Global Conference 2026, taking place on December 7 and 8, 2026, at ESPM’s Álvaro Alvim campus in São Paulo. We seek to discuss how artificial intelligence has been transforming communication and what challenges its use poses for information integrity, for voters’ ability to choose and for democracy in the 2026 elections.




