Brazil’s top electoral court has placed X’s recommendation system at the center of Brazil’s 2026 election campaign, demanding answers about a filter that may have made some candidates harder to discover while leaving others more widely promoted.

The immediate dispute is technical, but its political stakes are broad. The campaign of President Luiz Inácio Lula da Silva says X applied the restriction unevenly, creating an advantage for selected candidates. A right-leaning account of the case focuses more heavily on the court’s intervention, the short deadline imposed on the platform and the preservation of technical records for a possible audit. Both perspectives, however, describe the same central concern: election rules may have been applied inconsistently across competing campaigns.

What the court is demanding

André Mendonça, a minister of Brazil’s Superior Electoral Court, or TSE, gave X 24 hours to explain how the filter operates and which candidate accounts were excluded from the recommendation system. The court also ordered the company to preserve technical data, lists and code records connected to the tool, creating the basis for a possible judicial review.

The right-leaning magazine Oeste presents the order as a response to allegations from Lula’s campaign that X’s recommendation system had malfunctioned. Its account says the Workers’ Party claimed the filter applied to 665 accounts, while more than 1,500 candidates were left outside it. In that telling, the central issue is whether the platform’s rules were applied uniformly during an active campaign. ¹

The court’s demand for records does not itself establish that X deliberately favored or disadvantaged any candidate. It does, however, signal that the TSE considers the system’s operation important enough to preserve for a potential audit. The available reports do not include a response from X or a finding by the court that the platform violated election law.

The left’s argument: unequal amplification

The left-aligned outlet Brasil 247 frames the case more directly as an allegation of discriminatory algorithmic treatment. It says Lula’s legal team argued that X restricted algorithmic recommendations for only 665 candidacies, leaving more than 1,500 candidates outside the filter. According to that account, the result was that some profiles continued to receive amplification while others stopped being recommended. ²

The campaign’s requested remedy was sweeping: remove all candidate accounts from the recommendation filter for the duration of the election period. That proposal reflects the campaign’s broader concern that selectively limiting recommendations could change the relative reach of candidates even without removing their posts or suspending their accounts.

The distinction matters. A candidate whose account remains online but is no longer recommended may still publish content, yet reach fewer users through the platform’s discovery systems. Conversely, accounts that continue to benefit from recommendations could accumulate greater exposure over time. The reports describe this as a question of visibility rather than a conventional content-moderation dispute.

The right’s emphasis: process and evidence

The Oeste report gives greater weight to the mechanics of the court order. It highlights the 24-hour deadline, the requirement that X identify accounts excluded from the system and the preservation of data for a possible audit. Its framing is less an assertion that unequal treatment has already been proven than a description of an electoral authority seeking to determine how the tool was applied.

That distinction separates the two political perspectives. The left-aligned account starts with the campaign’s accusation that the filter produced an illegal imbalance and treats the court’s intervention as a response to that alleged harm. The right-aligned account foregrounds the judicial process and the unresolved technical question of whether the platform’s system treated candidates differently.

Neither perspective disputes the basic chronology. Lula’s campaign filed the complaint; Mendonça ordered X to provide explanations within 24 hours; and the platform was told to preserve information that could later be examined. The disagreement lies chiefly in emphasis: one side stresses the alleged political consequences of selective amplification, while the other stresses the need to establish what the algorithm actually did.

Why timing matters

Mendonça’s decision, as quoted in the reports, warns that the election campaign is already underway and that improper algorithmic recommendations could expand the reach of certain candidacies “progressively and in a way that is difficult to reverse.” ² That reasoning treats recommendation systems as cumulative. A small difference in visibility at one point in the campaign could become larger as users, followers and engagement build around the accounts receiving more exposure.

The court’s concern is therefore not limited to the existence of a filter. Platforms routinely use recommendation systems to sort and distribute content. The legal question raised in this case is whether a system intended to apply a common rule was used only for some candidates, and whether that difference altered the circulation of election information.

The reports do not explain why 665 accounts were affected, why more than 1,500 candidates were not, or whether the difference resulted from a technical error, an eligibility rule or another platform policy. Those are among the questions X has been ordered to address. The request for code records and technical lists suggests that the court wants evidence capable of distinguishing an accidental malfunction from a deliberate or systematic policy.

A broader test for platform neutrality

The dispute places X between competing demands. Campaigns want equal access to voters and may view opaque recommendation systems as a potential source of hidden advantage. Platforms, meanwhile, must explain complex systems that can change rapidly and may involve multiple eligibility rules or technical layers. The TSE’s order seeks to make those systems reviewable during the campaign rather than only after the election.

For Lula’s campaign, the reported filter is evidence that unequal digital visibility can affect electoral competition even when no account is formally blocked. For the court, the immediate objective is narrower: obtain an explanation, identify the affected profiles and preserve the underlying evidence. For the right-leaning account, that judicial scrutiny is the defining development; for the left-leaning account, the alleged imbalance is.

The next step depends on X’s response. Until the platform explains the filter and the court evaluates the preserved records, the available information supports an allegation of unequal treatment—not a final determination of wrongdoing. What is already clear is that Brazil’s election authorities regard algorithmic reach as part of the campaign environment, and that the neutrality of recommendation systems is now being tested in court.

---

2. TSE требует от X объяснений по поводу сбоев в алгоритме после жалобы кампании Лулы — Brasil 247 report on Lula’s campaign complaint, the alleged unequal amplification and André Mendonça’s warning about difficult-to-reverse reach effects.