Elections: Brazil

None

The TSE-AGU Agreement Explained

On Monday, 3 August 2026, the Superior Electoral Court (TSE) announced a cooperation agreement with the Democracy Observatory of the Attorney General’s Office (AGU) to confront the misuse of artificial intelligence in the coming elections. The TSE is the top body overseeing Brazil’s voting system.

The partnership, developed alongside the Electoral Judiciary School, will produce a handbook, known in Portuguese as a cartilha, setting out practical guidance for public agencies, political parties, candidates and the digital platforms that carry campaign content.

Officials say the document will gather recommendations to increase transparency around AI tools and reduce the risk of information manipulation during the campaign. Its release is expected later in August 2026.

What the Rules Would Actually Ban

The most prominent measure already mapped out is an express prohibition on deepfakes involving candidates or authorities. The guidance targets the digital manipulation of images and voices, and extends even to simulations of deceased personalities or fictional figures used to sway voters.

Campaigns would also be required to disclose when content was produced with the help of AI, so that voters can judge material accordingly. The rules bar automated amplification by bots or chatbots designed to inflate a message’s reach artificially.

A form of “technological silence” would prohibit distributing AI-produced or AI-modified content in the 72 hours before voting and the 24 hours after polls close. The material also warns about coordinated disinformation and so-called digital militias.

Why Deepfakes Are the Central Worry

A deepfake is synthetic audio or video that convincingly places words or actions onto a real person who never said or did them. In an election, a well-timed fake can spread faster than any correction, shaping perceptions before fact-checkers respond.

Brazil’s 2026 general elections are expected to be closely contested, raising the stakes for any tool that can distort a candidate’s message. Voice cloning is a particular concern because short audio clips are cheap to fake and easy to share on messaging apps.

The TSE has signaled since the 2024 municipal races that it wants clear limits rather than case-by-case improvisation. The new guidance builds on earlier resolutions that already required labeling of AI content and restricted the most deceptive uses.

The Bolsonaro Video That Sharpened the Debate

The push gained urgency after a video generated with AI circulated around the national convention of the Liberal Party (PL). The clip emulated former president Jair Bolsonaro appearing to voice an endorsement, illustrating exactly the kind of manipulation the rules aim to stop.

Episodes like that show why authorities want guardrails before the campaign peaks rather than after. A synthetic endorsement or a fabricated statement can move votes even if it is debunked within hours.

For the electoral court, the goal is to make the rules predictable so that parties know the boundaries and platforms know what to remove. Clarity, officials argue, is easier to enforce than a patchwork of emergency rulings.

How Brazil’s Approach Compares

Brazil has been among the more assertive democracies in regulating AI in elections, moving earlier than many peers to require disclosure and ban the most deceptive deepfakes. The country’s centralized electronic voting system gives the TSE broad authority to set campaign rules.

Elsewhere, many governments still rely on voluntary platform commitments or general disinformation laws rather than election-specific AI codes. Brazil’s model pairs binding electoral rules with a cooperation framework across state institutions.

The involvement of the AGU, which represents the federal government in court, signals that enforcement and legal defense of the measures are being planned together. That coordination could speed up responses when a viral fake appears mid-campaign.

What It Means for Voters and Foreign Observers

For Brazilian voters, the practical effect should be more labeling of AI content and faster takedowns of manipulated media. The rules do not ban AI outright; they aim to keep its use transparent and prevent impersonation of real people.

For foreign residents and observers, the framework is a useful signal of how Brazil intends to protect information integrity in a major election year. It also offers a reference point for other Latin American countries weighing similar steps.

The open question is enforcement. Rules on paper matter less than the speed at which platforms and courts can act once a convincing fake starts to spread across millions of phones in the final days of a campaign.

Frequently Asked Questions

What did Brazil’s TSE and AGU agree on?

On 3 August 2026 the TSE signed a cooperation agreement with the AGU’s Democracy Observatory to produce a handbook guiding the use of artificial intelligence by parties, candidates, public bodies and digital platforms in the 2026 elections.

What do the AI rules ban?

The guidance expressly prohibits deepfakes of candidates or authorities, requires disclosure of AI-generated content, bars automated boosting by bots, and forbids distributing AI content in the 72 hours before and 24 hours after the vote.

When do the rules take effect?

The handbook is expected to be published later in August 2026, ahead of the October 2026 general elections for president, Congress and state governors.

Sources: InfoMoney, Gazeta do Povo, Diário do Grande ABC.

This article was produced by The Rio Times’ automated newsroom system. How we use AI · Report an error