OpenAI denies ChatGPT post was related to Brazil’s election
🎧 Listen to this story
OpenAI denied to Tecnoblog that a post on ChatGPT’s global Instagram account had an electoral bias in Brazil. The post taught, in English, how to distinguish left from right using responses with more visual elements, and the company said its purpose was to present a new platform feature—not to comment on elections or political campaigns in the country.
Tecnoblog also reported that Meta faced criticism after Meta AI allowed the creation of election-related images for Flávio Bolsonaro but not for Lula. According to the site, Meta treated the case internally as an error in the AI’s limit settings; the company said the restriction followed Brazil’s election rules and announced additional measures to block prompts using only the candidates’ first names.
Why it matters · editorial interpretation
The episodes reported by Tecnoblog highlight how AI features can be interpreted in an electoral context and how limit settings affect requests involving candidates. The companies’ responses also indicate the importance of publicly clarifying whether a restriction resulted from an internal error or from election rules.
Sources
Society