Brazil elections face challenges from AI-driven content and platform algorithms

During the recent Brazilian electoral cycle, the impact of generative AI manifested differently than anticipated. While high-profile, single deepfake videos did not dominate the landscape, a high volume of smaller, distributed AI-generated materials circulated widely. Reports indicate that Agência Lupa cataloged 1,423 pieces of AI-modified content, including 520 deepfakes. Examples included an altered video of Flávio Bolsonaro and synthetic audio attributed to President Lula.

Beyond explicit deepfakes, significant concerns emerged regarding the “invisible” use of AI by digital platforms. Algorithms used by companies like Meta to manage feeds, ad segmentation, and content moderation were cited as potential sources of systemic bias. These systems can influence public discourse by modulating the visibility of specific candidates or campaign materials through ranking and recommendation logic.

The Superior Electoral Court (TSE) addressed these complexities through Resolution 23.755/2026, which prohibits deepfakes and requires the identification of AI-generated content. Crucially, the regulation also forbids AI systems from ranking, recommending, or prioritizing candidates, acknowledging that the risk extends from deceptive media to the underlying logic of content distribution.