AI-Generated Influencers Fuel Political Misinformation Surge in Brazil

A recent study revealed that Brazil hosts 18 artificial intelligence-generated political influencer profiles active between January 2025 and April 2026. These AI avatars have become significant sources of political misinformation, with 78% disseminating false claims about politicians and democratic institutions. The research, conducted by the Observatório das Eleições alongside Data Privacy Brasil and Aláfia Labs, highlights troubling trends ahead of Brazil’s upcoming elections.

Among these profiles, 61% fail to disclose their AI nature, making it difficult for users to discern their authenticity. Characters like 'Dona Maria,' an elderly black woman whose over 400 videos harshly criticize President Luiz Inácio Lula da Silva and the political left, represent a disturbing tactic to manipulate voter opinions. Meanwhile, 'Seu Zé da Feira,' an avatar portraying an elderly black man from a street market, vocally opposes right and centrist parties, labeling them as "a boss's union" indifferent to the people.

These virtual influencers primarily operate on TikTok and Instagram, with six profiles on each platform, supplemented by YouTube and others such as X, Kwai, and Facebook. Their disinformation campaigns often target prominent figures, including President Lula, former President Jair Bolsonaro, and Supreme Court justices Alexandre de Moraes, Cármen Lúcia, and Luís Roberto Barroso.

The research also exposed attempts to weaponize these avatars politically. For instance, the Brazil of Hope Federation (PT, PV, and PCdoB) requested the suspension of accounts linked to 'Dona Maria' at the Superior Electoral Court, alleging misinformation and premature electoral propaganda. Pro-government groups likewise created versions of this character with a pro-Lula narrative, demonstrating the strategic manipulation of synthetic personas across the political spectrum.

Experts underscored the challenge these AI-generated influencers pose to Brazil’s information environment. Despite technical flaws detectable upon close examination, these avatars often appear convincingly human to everyday users, simulating spontaneous political opinions and influencing debates during a critical election period.