Brazilian court gives Meta 24 hours to prove AI neutrality over looming presidential runoff

Meta denies giving preferential treatment to any candidate, says its rules apply equally to all political figures

A Brazilian court gave US tech company Meta 24 hours to prove that its artificial intelligence systems do not favour either candidate ahead of next Sunday’s presidential runoff, local media reported on Monday.

The Superior Electoral Court order, signed by magistrate Marco Aurelio Bellizze Oliveira, covers Meta's algorithmic systems managing social media platforms Instagram, Facebook, and WhatsApp, reported Brazilian outlet Universo Online.

The company must prove that it is not limiting the distribution of political content after President Luiz Inacio Lula da Silva's campaign alleged that it suppressed material favourable to the incumbent while permitting the spread of hostile AI-generated posts.

Read: Brazil's Lula accuses Meta of election interference

"These inconsistencies in metrics demand explanations and immediate correction," said Michel Bertoni, a lawyer representing Lula’s campaign.

Bertoni argued that when a major digital platform restricts the circulation of political content unevenly, it directly compromises electoral regularity and interferes with national sovereignty.

The measure follows reporting by Brazilian daily Folha de S.Paulo, which cited internal communications indicating that Meta employees acknowledged restrictions affecting Lula's accounts while algorithms recommended the profile and WhatsApp channel of challenger Flavio Bolsonaro.

Read More: AI able to manipulate elections worldwide, expert says

Meta denied giving preferential treatment to any candidate ahead of the October 25 runoff vote, maintaining that its rules apply equally to all political figures and that internal discussions focused on the severity of public technical issues rather than political bias.