FACTUM
Português

CASE #0028 · OPENED 5 NOV · UPDATED 2H AGO · CLOSED

Fake 'independent expert report' misuses TSE data to claim fraud in Brazil's 2026 presidential runoff

A document with hundreds of thousands of views on X says the extended voting hours flipped the runoff; the TSE data cited by Agência Lupa shows otherwise.

Evidence board

  • FalseAbout 1.8 to 2.5 million irregular votes were recorded during the two-hour extension of voting until 7pm.2 / 2 sources

    Agência Lupa cites TSE data showing only 34,393 votes were recorded after 5pm, about 0.027% of the total. That is far below the 2.5 million figure. Lupa also reports that a machine keeps receiving votes after 5pm only because voters are still in line, not because extra votes are being entered.

  • FalseIf only votes counted up to 5pm were considered, Flávio Bolsonaro would have won the runoff.2 / 2 sources

    According to Lupa, the votes recorded after 5pm (34,393) are too few to erase Lula's lead of 4.43 million votes over Flávio Bolsonaro.

  • FalseIn São Paulo, the 1,571 polling sections still open at 7pm gave Lula 57.46%, contrary to the rest of the state.1 / 2 sources

    The TSE told Lupa it is not possible to identify how people who voted after 5pm cast their ballots. The bulletin (BU) data the document says it used only gives each candidate's total for the whole day, so the percentage cannot be derived from it.

  • FalseVoters enabled without biometrics (poll worker release by typing birth year) are a material vulnerability that allows irregular votes.1 / 2 sources

    Lupa says enabling a voter without biometrics does not mean an irregular vote. According to the TSE, the voter must be present and show photo ID, and the birth year must match the electoral registry. The operation is logged on the machine and can be audited, and poll workers cannot vote in the voter's place.

  • UnclearThe 2026 data show extreme anomalies: near-unanimous sections concentrated in rural areas.1 / 2 sources

    The text of the Lupa article available to us is cut off right after this claim, so we cannot see how it was checked. We cannot judge it from the provided material.

Analysis

A document titled "Relatório Pericial Independente — Eleições 2026" is circulating on Instagram and X, where one version has passed 712,000 views. It claims that public data from Brazil's electoral court, the TSE, reveal fraud in the 2026 presidential runoff. Agência Lupa checked it and rated its central claims false.

What's being claimed

According to Lupa, the document makes several claims:

  • If only votes counted up to 5pm were considered, Flávio Bolsonaro (PL) would have won.
  • Between roughly 1.8 and 2.5 million "irregular" votes were recorded in the "extension" of voting until 7pm.
  • In São Paulo, 1,571 sections still open at 7pm gave Lula 57.46%, unlike the rest of the state.
  • Voters released by a poll worker without biometrics, by typing a birth year, are the biggest material vulnerability of the process.
  • The 2026 data show near-unanimous sections concentrated in rural areas.

The document is anonymous. Lupa calls it "apócrifo" (of unknown authorship), and nothing in the material provided identifies who wrote it or what "independent" body stands behind it.

What the evidence shows

The extension. Lupa reports that the two extra hours of voting were ordered by Supreme Court (STF) minister Cristiano Zanin, drawn as rapporteur of the case. Using TSE data, it says only 34,393 votes were recorded after 5pm, about 0.027% of the total. Lula's lead over Flávio Bolsonaro was 4.43 million votes, so that number cannot change the outcome. The claim of 2.5 million votes in the extension is therefore contradicted by the same public data the document cites.

The São Paulo percentage. The TSE told Lupa it is impossible to know how people who voted after 5pm voted. The CSV file with the bulletin (BU) data, which is the source the document claims to use, only lists each candidate's total for the whole day. A figure like 57.46% for late-closing sections cannot come from that file.

Biometrics. Lupa explains that voting without biometric validation is not an irregular vote. The voter must be present and show official photo ID, and the poll worker can only enable the machine if the birth year the voter states matches the registry. The TSE says the birth year is not in the voting book, so it cannot be easily read by the poll worker. Both the operation and who released the voter are recorded in the machine's log and can be audited. Voting or trying to vote in someone else's place carries up to three years in prison under article 309 of the Electoral Code. The TSE also notes that the Digital Vote Registry (RDV), in use since 2004, allows recounts while keeping the vote secret.

Rural sections. The Lupa text we have is cut off before this claim is addressed, so we do not rate it.

How outlets covered it

Agência Lupa, a fact-checking outlet, gives the detailed analysis. It quotes the document's passages, then answers each with TSE data and explanations. It says readers sent the content for checking via WhatsApp.

TV Sim Brasil carried a headline reaching the same conclusion, calling the report false and saying it uses TSE data to allege fraud. Only the headline was available to us, so we cannot say more about its reporting.

The two outlets agree on the verdict. We found no outlet in this set defending the document, and we did not see the document itself, only Lupa's quotes from it.

What we still don't know

  • Who wrote the document and who first spread it.
  • How Lupa assessed the rural-section claim, since the text is cut off.
  • The details of the court case in which Zanin was drawn rapporteur and ordered the extension. The Lupa excerpt mentions it but does not describe it.
  • Whether any official audit or recount has been requested. Neither source mentions one.

What the available evidence does show is that the document's main numbers do not match the TSE's public data, and its claims about the biometric procedure ignore the safeguards the TSE describes.

Sources

Comments