AI chatbots sway political information in Brazil and Portugal

A study by Brazil’s Institute of Technology and Society (ITS Rio) examined how seven leading AI chatbots – ChatGPT, Gemini, Meta AI, DeepSeek, Grok, Perplexity and Claude – responded to questions about the 2026 elections in five states. Eighty‑five percent of the tools ranked or recommended presidential and state candidates, breaching the Brazilian Electoral Court’s Resolution 23.755/2026 that bans AI from influencing voter choices. Only Meta AI complied with the rules. Researchers noted that 54 % of Brazilians used at least one generative AI service in 2024, and 62.9 % said they might consult an AI for candidate information, underscoring the technology’s growing role in shaping electoral perceptions.

A separate report from OberCom and the Reuters Institute surveyed Portuguese internet users about AI‑generated news. Just 24 % expressed confidence in information obtained from chatbots, compared with 40 % for search engines and 51 % for mainstream news. Respondents said chatbots are mainly used to summarise or clarify content rather than replace traditional journalism. The study also highlighted the rising influence of digital creators and influencers among younger audiences, while noting that only 8 % of Portuguese have paid for online news in the past year.