Internal communications among Meta employees confirm that Instagram accounts belonging to Brazilian President Luiz Inacio Lula da Silva were negatively affected by platform restrictions during the country's ongoing presidential election. The messages, reviewed by Brazilian daily Folha de S Paulo and reported Friday, also acknowledge that Meta's platforms recommended the profile and WhatsApp channel of Lula's right-wing challenger, Senator Flavio Bolsonaro. Meta's defense — that these were pre-existing technical issues and the internal discussions concerned severity, not political interference — is doing a lot of heavy lifting. The timing is brutal for Meta. Lula and Bolsonaro face a presidential runoff on October 25, following Bolsonaro's stronger-than-expected first-round performance. Lula's party documented specific degradations: posts appearing without share buttons, reduced reach, and missing options for reactions, audio, and playback speed. These are not abstract algorithmic complaints — they are measurable, discrete feature removals that directly suppress a candidate's ability to campaign on the platform. A day before the Folha report dropped, a judge on Brazil's top electoral court had already acted. Justice Marco Aurelio Bellizze ruled Thursday that Meta cannot reduce political content distribution, restrict searches, or remove posts in ways that create imbalance between the two candidates. The ruling mandates identical AI criteria for both candidacies and carries a fine of 500,000 reais ($98,700) per day for noncompliance. Meta's public response — "Any suggestion that Meta favors one side is false" — is a blanket denial that the internal communications appear to contradict. The company told Reuters the technical issues were already public, framing the leaked messages as merely discussing severity. But acknowledging that one candidate's features were degraded while the other's profile was being recommended is not a severity discussion — it is an asymmetry discussion. The structural question is who controls the information architecture of democratic elections in the Global South. Brazil has 130 million Instagram users and WhatsApp is the country's dominant communications channel. When a single company's algorithmic choices can suppress share buttons on one candidate's posts while recommending another's channel, the platform is not neutral infrastructure — it is an unaccountable electoral actor. The court's intervention recognizes this, but enforcement against a $1.5 trillion company via $98,700 daily fines is a mismatch of several orders of magnitude. Brazil is emerging as the most aggressive jurisdiction in the world on platform accountability during elections, but the tools available — post-hoc court orders, modest fines, reactive complaints — are structurally inadequate. The asymmetry is always temporal: algorithmic suppression happens in real time during a campaign, while legal remedies arrive days or weeks later. Even if Meta complies perfectly from this moment forward, the first-round vote already happened under the contested conditions. The 20-year trajectory here is clear. If platform companies retain unilateral control over political content distribution with no pre-election transparency requirements, algorithmic interference — whether intentional or through convenient "technical issues" — becomes a permanent feature of elections in countries where these platforms dominate the information ecosystem.
