Algorithms should not be allowed to shape electoral choice
As Brazil prepares for its October presidential election, the Superior Electoral Court is confronting a question that democracies worldwide are only beginning to reckon with: how to preserve the integrity of human choice in an age when algorithms can quietly shape it. Under newly installed president Nunes Marques, the court is not merely issuing prohibitions but constructing lasting institutional architecture — commissions, regional security units, academic partnerships — to meet the challenge of AI-driven disinformation with something more durable than reaction.
- Manipulated content and AI-generated disinformation have spread rapidly across Brazil's digital networks, raising alarm among electoral officials with October's presidential vote drawing near.
- A permanent commission has been given just 90 days to produce a national catalog of solutions against AI-driven electoral crimes — a compressed timeline that underscores how seriously the court views the threat.
- Regional electoral courts across the country must stand up their own information security units within 30 days, weaving a distributed surveillance network into Brazil's electoral infrastructure.
- The court has already banned AI providers from suggesting candidates to voters — even when asked — drawing a firm line between algorithmic influence and voter autonomy.
- Nunes Marques plans to meet directly with political parties to reinforce compliance, signaling that enforcement, not just regulation, is the court's intent.
Two weeks into his tenure as president of Brazil's Superior Electoral Court, Nunes Marques convened the heads of the country's regional electoral tribunals and made his priorities clear. With October's presidential election approaching and AI-driven disinformation spreading across digital networks, the court would not wait for the threat to materialize further — it would build the institutions needed to confront it.
At the center of the court's response is a permanent commission charged with developing a national catalog of solutions to combat AI-driven electoral crimes and content manipulation. The commission has 90 days to complete its work and will draw on partnerships with universities specializing in digital crime investigation. In parallel, regional electoral courts have been ordered to establish their own information security units within 30 days, creating a distributed layer of cyber vigilance across the country's electoral system.
The court's concern with artificial intelligence predates this announcement. In March, it approved restrictions on AI use in campaign propaganda and, most strikingly, banned AI providers from suggesting candidates to voters — even when users explicitly asked for such guidance. The reasoning was principled: algorithms should not be permitted to shape electoral choice, regardless of how the request is framed.
What sets the May announcement apart is its institutional ambition. Rather than relying on prohibitions alone, the court is investing in the capacity to detect violations and respond to them. Marques has also committed to meeting with political parties directly to reinforce adherence to electoral rules — a signal that compliance will be expected, not merely encouraged. Whether the measures will prove equal to the challenge remains open, but the court is treating AI-driven interference as a problem that demands not just rules, but enduring institutional resolve.
Brazil's electoral authorities are moving to fortify the country's presidential vote against artificial intelligence manipulation. On May 25th, Nunes Marques, who took over as president of the Superior Electoral Court just two weeks earlier, announced a series of safeguards designed to prevent the misuse of AI during October's presidential election. The announcement came during his first meeting with the heads of Brazil's regional electoral tribunals, signaling that election security would be a defining priority of his tenure.
The court's concerns are concrete. In recent months, manipulated content and disinformation have proliferated across digital networks in Brazil, creating what electoral officials view as a genuine threat to the integrity of the voting process. Rather than respond with ad-hoc measures, the court is building institutional infrastructure to meet the challenge. The centerpiece is a permanent commission tasked with developing a national catalog of solutions to combat AI-driven electoral crimes and content manipulation. This commission has 90 days to complete its work—a tight timeline that reflects the urgency officials feel as October approaches.
The commission will not operate in isolation. It will coordinate with universities that specialize in investigating digital crimes, drawing on academic expertise to inform its recommendations. Simultaneously, Brazil's regional electoral courts have been ordered to establish their own information security units within 30 days. These units will focus on strengthening data protection and monitoring for cyber threats, creating a distributed network of vigilance across the country's electoral infrastructure.
Marques has also committed to meeting with political parties to reinforce the importance of adhering to existing electoral rules during their campaigns. This signals that the court intends to enforce compliance, not merely suggest it. The court's approach reflects a recognition that AI-driven threats to elections are not purely technical problems—they involve human actors making choices about how to deploy these tools.
The court's concerns about AI are not new. In March, it approved specific restrictions on artificial intelligence use in electoral propaganda and campaign activities. Most notably, the court banned AI providers from offering voters suggestions about which candidates to support, even when users explicitly request such guidance. The reasoning is straightforward: algorithms should not be allowed to shape electoral choice. The court views this as essential to preserving the voter's autonomy and, by extension, the legitimacy of the democratic process itself.
What distinguishes the May announcement is its scope and institutional weight. Rather than issuing prohibitions and hoping they stick, the court is building the capacity to detect violations, investigate them, and respond. The 90-day timeline for the permanent commission suggests the court expects to have concrete recommendations in place well before the October election. Whether these measures will prove sufficient remains to be seen, but the court is clearly treating AI-driven electoral interference as a problem that demands sustained institutional attention.
Citações Notáveis
The court views algorithmic interference in candidate selection as a threat to voter autonomy and democratic legitimacy— Brazil's Superior Electoral Court position on AI restrictions
A Conversa do Hearth Outra perspectiva sobre a história
Why did the court feel compelled to act now, in May, when the election isn't until October?
Because disinformation and manipulated content have already been spreading. The court isn't waiting for the problem to peak—it's trying to get ahead of it.
The 90-day timeline for the commission seems tight. What happens if they don't finish?
That's the real question. The court is signaling that this is urgent, but 90 days to develop a national strategy is ambitious. They may have to work in phases.
Why involve universities in this? Why not just rely on the court's own technical staff?
Universities have forensic expertise in digital crimes that the court may not have in-house. It's an acknowledgment that this problem is too big and too specialized for any single institution to solve alone.
The ban on AI suggesting candidates—how would that even be enforced?
That's where the regional security units come in. They'll be monitoring for violations, though detecting subtle manipulation is harder than detecting an outright ban.
Does this suggest the court is worried about a specific threat, or is this more general?
It's both. The court is responding to what's already happening—disinformation campaigns—but also trying to prevent worse scenarios before they materialize.
What's the political risk here? Could parties claim the court is overreaching?
Possibly. But the court is framing this as protecting electoral integrity, not favoring any particular side. The real test will be whether enforcement is even-handed.