Brazil's Tribunal Superior Eleitoral (TSE) is proposing a "seal of accuracy" for electoral polls to ensure the reliability of polling methodologies [1].
This move comes as the court seeks to standardize how public opinion is measured during a volatile election cycle. The effort to regulate polls aims to prevent the spread of misinformation and ensure that the data influencing voters is based on sound scientific practices.
On June 10, 2026, the TSE discussed these new rules with polling institutes [2]. The scale of the current electoral landscape is significant, with more than 700 electoral polls registered for 2026 [2]. These activities have moved approximately R$ 40 million [2].
Parallel to these regulatory discussions, the court has faced specific legal challenges regarding poll results. On June 9, 2026, the TSE postponed a decision regarding the suspension of a poll conducted by AtlasIntel [1]. The poll in question had been suspended following a request from Flávio Bolsonaro after the data indicated a decline in his support [3].
The postponement occurred after a request for review interrupted the judgment process [4]. The court is now weighing whether the suspension of the AtlasIntel poll was justified or if it interferes with the freedom of data collection.
By implementing a seal of accuracy, the TSE intends to create a verifiable benchmark for institutes. This system would theoretically allow voters to distinguish between rigorous scientific surveys and those with flawed methodologies, a critical distinction in a high-stakes election year.
“The TSE is proposing a “seal of accuracy” for electoral polls.”
The TSE's push for a certification system reflects a growing tension between political figures and data providers in Brazil. By attempting to institutionalize 'accuracy,' the court is moving from a passive registry of polls to an active validator of methodology. This could either increase public trust in electoral data or create a new legal battleground over who defines scientific validity in political surveying.
