Publication date: 16 October 2026
Brazil’s presidential runoff has produced a consequential test of how far an electoral court can go when a campaign alleges that a platform’s automated systems are distorting political competition. A judge on the Superior Electoral Court ordered Meta to take immediate measures to provide equal treatment to President Luiz Inácio Lula da Silva and challenger Flávio Bolsonaro ahead of the 25 October vote.
The order followed a complaint from Lula’s campaign alleging that content associated with the incumbent suffered unexplained reductions in reach and that Meta’s artificial-intelligence tools treated the two candidacies differently. Meta has rejected accusations that its systems favour Bolsonaro, attributed reported problems to technical glitches affecting users beyond Brazil and said it complies with Brazilian election rules. The court’s intervention is therefore a safeguard imposed while disputed technical and political claims remain unresolved—not a final proof of intentional bias.
From neutral rules to equal effects
The decision is notable because it focuses on parity in the operation of the platform. According to Associated Press reporting, the order says Meta must not reduce distribution, restrict searches or remove posts in a way that creates an imbalance, and that its AI tool must apply identical criteria to both campaigns. Non-compliance carries a daily fine of 500,000 reais.
That creates a demanding governance problem. Platforms can publish formally neutral policies, yet candidates may still experience different outcomes because of automated ranking, content classifiers, advertising systems or product failures. An electoral authority seeking “equal treatment” must decide whether it is testing the rules, the engineering process or measurable reach. Those are not equivalent.
Brazil’s electoral court has also been pressing Meta on disinformation, the misuse of AI and preservation of WhatsApp-channel metadata when campaign content becomes evidence in election cases. Together, those actions show a court trying to make fast-moving platform infrastructure legible during a short runoff campaign.
Transparency is the missing bridge
Courts can order parity more quickly than they can independently audit a recommendation system. Reach varies for many legitimate reasons, including audience size, engagement, timing and content format. Requiring identical political outcomes would be unrealistic; requiring consistent enforcement, documented testing and rapid correction of failures is more defensible.
The most durable response would therefore be procedural. Platforms should preserve relevant logs, disclose material incidents to electoral authorities, test whether comparable campaign content receives comparable treatment and provide a route for urgent review. Courts, in turn, should distinguish between evidence of discriminatory enforcement and ordinary variation in audience response.
The wider leadership lesson extends beyond Brazil. Election integrity increasingly depends on systems owned by private companies but used as public communications infrastructure. Democratic institutions cannot simply assume those systems are neutral, while governments must avoid converting fairness oversight into political control of distribution.
Brazil’s order places that tension in plain view. The immediate dispute concerns two candidates and one company. The precedent concerns whether algorithmic accountability becomes a routine part of administering elections.