Redcuban1959 [any]

The president of the Superior Electoral Court (TSE), Kassio Nunes Marques, determined this Monday (8) the suspension of the publication of an AtlasIntel survey on the 2026 presidential race. The decision partially responds to a request presented by Senator Flávio Bolsonaro (PL-RJ) and the party, who questioned the methodology used by the institute.

The poll, registered under number BR-06939/2026, indicated a drop in Flávio Bolsonaro’s voting intentions after the release of audio recordings of conversations between the senator and banker Daniel Vorcaro regarding the negotiation of approximately R$ 132 million for the film Dark Horse, about the political trajectory of former president Bolsonaro. The survey also indicated an increase in the pre-candidate’s disapproval rating.

In the survey, President Lula (PT), Flávio’s main opponent, appeared with 48.9% of voting intentions, compared to 41.8% for the PL senator. In the previous round, conducted in April, before the disclosure of the conversation between Flávio and Vorcaro, the parliamentarian had 47.8%, while Lula totaled 47.5%, indicating a negative fluctuation of six points for the senator.

Among those interviewed who stated they were aware of the leaked conversations between Flávio Bolsonaro and Vorcaro, 51.7% said they saw evidence of the senator’s direct involvement in the Banco Master scandal. Another 33.3% assessed that the conversations show a legitimate attempt to obtain financial support for the production of a film about Jair Bolsonaro. Meanwhile, 12.1% said they saw only a close relationship between Flávio and the bank owner, without proof of illegality.

When filing a complaint with the TSE (Superior Electoral Court) in May, Flávio Bolsonaro’s legal team said that the institute had induced respondents to associate Senator Flávio Bolsonaro with a negative perception by playing the audio of a conversation between him and businessman Daniel Vorcaro before the questions. Flávio’s defense classified the survey as a “serious manipulative precedent” and alleged that the format adopted by AtlasIntel could contaminate the respondents’ answers.

In the preliminary ruling, Nunes Marques prohibited AtlasIntel from promoting, boosting, republishing, or maintaining the survey on its official channels until further deliberation by the Court. The case will still be analyzed by the full TSE (Superior Electoral Court).

The PL argued that the questionnaire contained elements capable of influencing respondents. According to the complaint, the survey associated Flávio Bolsonaro’s name with the case involving Vorcaro and Banco Master before collecting opinions on voting intention, rejection, and electoral viability. The defense maintained that the sequence of questions could compromise the neutrality of the survey.

In granting the injunction, Nunes Marques stated that there was evidence of possible compromise in the methodology used by AtlasIntel. “The elements brought to the record after the defendant’s statement reinforce, in a summary judgment, the relevant indications of compromise in the methodology of the challenged research, including in comparison with the questionnaires of other surveys registered with the TSE by the same company,” the minister highlighted.

The decision also mentions statements made by AtlasIntel CEO Andrei Roman in an interview on May 19. According to the judge, the executive’s public statements about the potential electoral impact of the content presented to interviewees reinforced the doubts raised by the complaint.

At the time, Roman told the press that it was “clear that there is no immediate collapse of Flávio Bolsonaro.” At the same time, it was “quite clear that it will be difficult for him to recover his image and electoral viability going forward.” The CEO also said that “it is not the first time that an Atlas poll has been challenged in electoral court” and that “in several election cycles” the company “responds to these actions and to this day I don’t remember a case that we haven’t managed to win in the end.”

In a statement, AtlasIntel reports that it will respect the decision issued by Nunes Marques, but reaffirmed “the technical robustness and legality of the study.”

“It is important to emphasize that the research was conducted without the controversial audio being played for respondents during the questionnaire. The main questionnaire was fully completed and submitted before any contact between the participant and the audiovisual content. There was no inducement of the interviewees. The entire methodological design of the questionnaire, as well as the dynamics of the audio test application, was conducted with the technical and scientific rigor that characterizes AtlasIntel’s work, always guided by the principles of impartiality, transparency, methodological integrity, and statistical quality of the data produced.”

Andrei Roman stated in a publication on X that the recording had no impact on the electoral scenarios tested. “The audio is played after the completion of the survey questionnaire and therefore has no impact on the electoral scenarios. The idea is to understand in real time the impact of the audio on the electorate’s perception , with demographic segmentation. AtlasIntel always maintains an impartial stance, which characterizes our work not only in Brazil, but globally,” he declared in a publication on X.

In addition to suspending further releases of the survey, Nunes Marques ordered AtlasIntel to submit supplementary technical documentation to clarify aspects of the methodology used. The Electoral Public Prosecutor’s Office was also summoned to comment on the case.