Online abuse of candidates Study examines AI images

Date unknown

Ana Graziela Aguiar reports an InternetLab study with Instituto Democracia em Xeque and Núcleo Jornalismo analyzing more than 11,000 X and TikTok images and 15 candidacies across the political spectrum. The report gives misogyny in 25% of analyzed content, fatphobia in 14% and transphobia in 11%. Catharina Vilela says one in four offensive images used AI and emphasizes that the insulting depictions generally did not seek to appear real. She describes political deterrence and threats extending to families. The package recalls Meta's January-2025 moderation reduction, airs Mark Zuckerberg's justification and Julie Ricard's engagement-model critique. Meta says attacks based on protected characteristics are prohibited and violations removed. Each position and denominator remains attributed; no causal finding or new Meta decision is inferred.

TV Brasil describes an InternetLab, Democracia em Xeque and Núcleo MonitorA study examining more than 11,000 images involving fifteen women's candidacies in Brazil's 2026 election. It attributes 25% to misogyny, 14% to fatphobia and 11% to transphobia. Catharina Vilela says about a quarter of offensive images used AI, often without trying to look realistic, and describes intimidation of candidates and families. Julie Ricard criticizes platform moderation, while the report preserves Meta's response.