With Brazil’s high-stakes presidential runoff taking place today, October 25, 2026, a major controversy has erupted over allegations that Meta AI — the generative artificial intelligence tool built into Facebook, Instagram, and WhatsApp — discriminated against incumbent President Luiz Inácio Lula da Silva by refusing to generate images of him while producing campaign material for his rival, right-wing Senator Flávio Bolsonaro.
Lula’s campaign filed formal complaints alleging that prompts combining Lula’s first name with his ballot number, 13, were systematically refused by Meta AI, while the same pattern using “Flávio” and number 22 returned usable images of Bolsonaro. Brazil’s Superior Electoral Court has already ordered Meta not to interfere with political content distribution, and a judge requested technical explanations from both Meta and X (formerly Twitter).
Meta denied political bias, attributing the discrepancies to unrelated technical failures and stating that it had taken steps to restore equal treatment for both candidates by October 20. However, federal prosecutors have counted more than 1,000 cases of allegedly disproportionate content moderation in September and October alone. The United Nations and US lawmakers have also weighed in, demanding answers from the company in one of its most lucrative global markets.
Source: AI Weekly – Meta AI Faces Brazil Court Order Over Lula Image Refusal