Brazil’s Superior Electoral Court (TSE) has given the country’s major digital platforms until Sunday, August 16, to file concrete plans for preventing disinformation and other electoral abuses during the 2026 general election.
The directive, issued in a portaria signed by TSE President Kassio Nunes Marques in late July, applies to any service with more than five million users in Brazil. It covers Facebook, Instagram, TikTok, X (formerly Twitter), WhatsApp and Telegram.
Each company must disclose the organisational structure it will use to receive and implement rulings from the electoral judiciary, the monitoring mechanisms it will employ and the methodology for collecting and delivering data to the TSE.
Additionally, the platforms must explain how they will halt the boosting and monetisation of content deemed false or severely decontextualised when such material threatens the integrity of the vote.
Coordinated inauthentic behaviour
The order devotes a specific section to so‑called coordinated inauthentic behaviour, in which groups of accounts, fake profiles or automated systems act in concert to manipulate public debate.
Platforms are required to state the criteria they will use to detect such networks and describe the steps they will take to interrupt, limit or hinder their activity.
Any detected instances must be reported to the TSE, preserving information that could be relevant for future investigations.
Public and restricted information
The filings will be split into a public portion, accessible to anyone, and a restricted portion containing technical data deemed sensitive.
Access to the restricted data will be limited to TSE officials and staff formally designated by the court.
Companies may, exceptionally, withhold certain restricted items, but they must justify the decision on legal grounds such as industrial or commercial secrecy, technical infeasibility or conflicting rulings from other jurisdictions.
The TSE president will review these justifications and may request further clarification, order adjustments to the submitted plans or conduct technical audits.
Artificial intelligence safeguards
Platforms that offer generative artificial intelligence systems — language models, chatbots and automated assistants — must detail the protections they have adopted to prevent misuse of these tools during the campaign.
They must demonstrate how they will stop the systems from producing content that favours any candidate, party, federation or coalition, or that provides voting recommendations.
Operators also have to explain how they will block the generation of non‑consensual sexual content involving candidates and electoral propaganda that constitutes gender‑based political violence.
Furthermore, firms need to disclose the existence and frequency of tests performed on their AI systems before and during the election period to verify the tools’ resistance to attempts to bypass the safeguards.
The results of those tests, however, will remain in the TSE’s restricted access area.