AI-Generated Attacks Target Female Candidates in Brazil Election

Artificial intelligence has emerged as a major weapon for online harassment in Brazil’s 2026 general election campaign. A newly released study by MonitorA, an observatory that tracks gender-based political violence, revealed that AI was used to generate 25% of all offensive images and insults targeting female candidates on the social media platform X. The findings highlight the steep challenge facing South America’s largest democracy as it attempts to police synthetic media under some of the world's strictest electoral rules.

The study, conducted by internet policy group InternetLab and feminist magazine AzMina, analyzed 11,137 images published in comments on the profiles of 132 female candidates. Researchers identified 660 offensive images, finding that 165 of them bore clear signs of AI generation or manipulation. Rather than using high-quality "deepfakes" designed to trick voters into believing a scenario is real, bad actors are primarily using AI to mass-produce cheap memes, caricatures, and altered photos designed to humiliate, dehumanize, and mock candidates.

Prominent progressive politicians bore the brunt of the digital onslaught, with three candidates concentrating nearly half of the identified offensive images. Environment Minister Marina Silva, federal deputy Sâmia Bomfim, and Erika Hilton—one of Brazil's first openly transgender federal lawmakers—were the primary targets. The nature of the AI attacks varied by target: Sâmia Bomfim faced a barrage of body-shaming, fatphobic alterations; Erika Hilton was targeted with racist and transphobic imagery; and Marina Silva was depicted with ageist physical distortions and compared to animals.

The surge in synthetic harassment comes despite aggressive regulatory efforts by Brazil's Superior Electoral Court (TSE), the federal institution that oversees the country's voting process. Ahead of the 2026 vote, the TSE enacted strict regulations completely banning deepfakes in campaign materials and requiring explicit, transparent labeling of any AI-generated content. Under these rules, candidates who use unauthorized or deceptive AI can face the revocation of their registration or mandate.

However, enforcing these rules is proving difficult because the vast majority of the offensive content does not come from official campaigns, but from anonymous accounts, coordinated bot networks, and ordinary social media users. While the TSE has established channels to notify major tech platforms to remove non-compliant content, the sheer volume of decentralized, AI-generated memes makes real-time moderation on platforms like X incredibly difficult.

As Brazil heads toward the final stretch of the election cycle, the weaponization of low-cost generative AI tools is raising alarms among democratic watchdogs. Experts warn that the targeted nature of these digital campaigns could further discourage women and minority candidates from entering public life in a country where female representation in Congress still hovers around 19%, among the lowest in Latin America.