Eleições 2026: pesquisa revela que 20% das respostas de IAs recomendaram candidatos
2026-09-25
Portal O Dia
The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.
A study by ITS Rio and the Federal Public Ministry found that in August 2026, 20% of responses from popular AI tools in Brazil recommended political candidates, violating electoral regulations. Perplexity, Meta AI, Grok, and DeepSeek were the main offenders, raising concerns about AI's influence on democratic processes.[AI generated]
Why's our monitor labelling this an incident or hazard?
The article explicitly states that AI systems produced responses recommending candidates and ranking them, which is prohibited by law. This use of AI has directly influenced electoral information presented to voters, potentially affecting election fairness and violating legal frameworks protecting political rights. The AI systems' outputs have caused realized harm by breaching electoral regulations and possibly impacting democratic processes, which fits the definition of an AI Incident involving violations of human rights or legal obligations. The presence of hallucinations and outdated information further supports the occurrence of harm through misinformation. Therefore, the event is best classified as an AI Incident.[AI generated]