The 2026 Brazilian Elections: A Perfect Storm of Artificial Intelligence, Disinformation, and Regulatory Gaps
As Brazil’s 2026 electoral campaign officially commenced on Sunday, the nation finds itself at a critical juncture where the convergence of artificial intelligence, social media platforms, and a regulatory framework struggling to keep pace threatens to fundamentally reshape the democratic process. The upcoming elections represent the first major electoral contest in which AI is not merely an emerging technology but a firmly established, accessible tool available to virtually any user with an internet connection. Unlike previous elections where the use of such technology was limited by cost and technical expertise, the current landscape presents a dramatically different scenario—one where the barriers to entry have collapsed, making sophisticated disinformation campaigns accessible to anyone, regardless of their technical proficiency or financial resources. Débora Salles, the general coordinator of NetLab, a research laboratory at the Federal University of Rio de Janeiro’s School of Communication, has been studying the social impact of the internet and social media since 2013, and her recent research paints a troubling picture of what lies ahead. The problems that plagued the 2022 and 2024 elections—mass disinformation, foreign interference, and coordinated inauthentic behavior—have not only persisted but have been amplified exponentially by the new technological capabilities. At the heart of this concern lies a fundamental structural weakness: the electoral justice system appears unprepared for the scale of the challenge, and platforms have demonstrated little genuine commitment to ensuring the integrity of the electoral debate.
The issue of illegal and irregular political advertising has emerged as a particularly concerning aspect of Brazil’s evolving electoral landscape, with researchers identifying a troubling pattern of foreign interference that appears to be growing more sophisticated and more difficult to trace. During the 2024 municipal elections, NetLab documented 137 illegal advertisements circulating on social media platforms, with a significant portion of these ads linked to accounts registered in other countries, including Portugal, the United Kingdom, and Belgium. Perhaps most alarming was the discovery of advertisements promoting Pablo Marçal, then a candidate for mayor of São Paulo, that were being run through TikTok pages based outside Brazil, with sources traced to countries across Europe, Asia, and the United States. This international dimension to electoral interference represents a serious challenge to Brazil’s electoral sovereignty, as it allows external actors to influence domestic elections while remaining largely beyond the reach of Brazilian authorities. The platforms, for their part, have shown little initiative in addressing these vulnerabilities, with Meta, for example, continuing to allow pages from China to publish fraudulent advertisements about public policies and ads originating from Colombia that discuss Brazilian candidates. The fundamental problem, according to Salles and her colleagues, is that platforms operate in an opaque environment where transparency is minimal and effective monitoring is virtually impossible. Despite electoral rules explicitly prohibiting foreign advertising in Brazilian elections, there exists no robust mechanism to enforce these regulations, as platforms have failed to update their verification systems to adequately identify the true origins of political advertisements. This creates a situation where the stated rules and the actual practices on the ground are fundamentally disconnected, with violations occurring with impunity and no meaningful consequences for those who perpetrate them.
The introduction of sophisticated artificial intelligence capabilities into this already compromised ecosystem has dramatically escalated the threats facing Brazilian democracy, creating what researchers describe as a “perfect storm” of technological capability and regulatory inadequacy. The production of manipulated content has become extraordinarily cheap and easy, requiring no specialized skills or significant financial resources to create convincing deepfakes and synthetic media that can be deployed to deceive voters or attack political opponents. Salles emphasizes that all the problems that existed in previous elections remain fully present, but they now possess a “booster” that delivers unprecedented scale at an affordable cost, fundamentally changing the nature and magnitude of the threat. The economic incentives inherent in social media platforms’ business models compound this problem, as AI-generated content tends to generate significant engagement, which in turn produces greater financial returns for the companies that host it. This creates a troubling dynamic where platforms have an economic interest in allowing problematic content to circulate, as sensational and emotionally charged material tends to drive user interaction and ad revenue. The recent video featuring Jair Bolsonaro that was shown at Flávio Bolsonaro’s convention exemplifies these concerns, with researchers identifying it as a deepfake based on TSE resolution guidelines, which explicitly prohibit the use of fabricated or manipulated content in election campaigns. The resolution is unambiguous: synthetic content cannot be used to benefit one candidacy or harm another, and fabricated content purporting to show real individuals is explicitly prohibited. However, the gap between the letter of the law and its enforcement remains vast, with the technological capability to create such content far outpacing the ability of authorities to detect and respond to violations in a timely manner.
The Superior Electoral Court’s response to these challenges has been characterized by a mix of effort and insufficiency, as the institution attempts to update its regulatory framework while facing the inherent limitations of enforcement in a digital environment. Two weeks before the campaign officially began, the TSE issued a guide with recommended practices for the use of AI in elections and updated its rules to address synthetic content, representing an acknowledgment of the evolving threat landscape. However, researchers argue that these measures, while commendable in intent, remain fundamentally inadequate for addressing the scale of the challenge. The rules place significant responsibility on campaigns rather than platforms, even though platforms are the primary vehicles for content amplification and distribution. When fines are imposed on platforms for violations, they tend to be relatively low, and companies have considerable room to appeal decisions, often leading to proceedings that drag on in court with little progress. The TSE’s relationships with technology companies are described as “almost generous,” based on principles of cooperation rather than obligation, allowing platforms to volunteer their assistance without being held to meaningful accountability standards. In July, the TSE took the initiative to send notices to regional electoral courts authorizing partnerships to monitor synthetic content, an acknowledgment that the central institution cannot effectively monitor the entire landscape alone. However, Salles questions whether sufficient time remains to build an effective oversight infrastructure for an election in a country of continental dimensions, noting that such an initiative would have been better started months ago. The fundamental challenge lies in the fact that Brazil’s elections will be the first large-scale elections in which AI is already accessible and widespread, meaning there is a genuine lack of precedent for how to address the emerging challenges. This uncertainty is compounded by the significant lobbying presence of social media platforms and AI companies, which actively work to prevent regulations that might interfere with their operations or reduce their profitability.
The broader regulatory environment in Brazil has created a situation where the courts have become the last resort for addressing online harms, yet the absence of comprehensive legislation leaves the system struggling to respond effectively. Salles argues that ideally, a legislative process would need to approve rules governing these platforms, establishing clear standards for transparency, accountability, and enforcement. However, the current political environment makes such legislative action difficult to achieve, leaving the executive branch to address issues through decrees that respond to illegal content and other problems but lack the enforcement mechanisms necessary to be truly effective. This regulatory vacuum places an enormous burden on the judicial system, which must attempt to enforce rules without adequate tools or support. The fundamental precondition for meaningful regulation is the ability to monitor what actually happens on social media platforms, yet government bodies, civil society organizations, and academic researchers lack the access and technical capabilities to conduct such oversight effectively. Without this basic monitoring capacity, holding platforms accountable becomes virtually impossible, and the stated rules exist primarily as aspirations rather than enforceable standards. The responsibility this creates for individual voters is troubling to researchers, who are reluctant to place the burden of identifying misinformation on ordinary citizens who lack the tools or expertise to make such determinations reliably. Salles advises voters to check whether reputable news organizations have covered the political content they encounter and to verify whether such content plays a legitimate role in public debate. The TSE has launched a system for reporting problematic content, and while reports filed directly with social networks often lead nowhere, Salles encourages their use as a mechanism for helping the electoral court identify and respond to violations. The sheer volume of information being shared makes it impossible for any single organization to monitor everything, meaning public assistance with monitoring is essential to the TSE’s effectiveness, even if individual reports may not result in immediate consequences.
Looking ahead to the campaign that has just begun, researchers express profound concerns about the challenges that await Brazilian democracy, while simultaneously acknowledging the opportunities for public engagement and pressure that can help shape responses to these emerging threats. Salles believes that Brazilian society is facing a challenge of enormous proportions, and the electoral system does not appear ready for what is coming. There are no clear answers about how to confront the challenges posed by AI-enabled disinformation or whether the measures already in place will prove effective in protecting the integrity of the electoral process. However, there remains the possibility of identifying problems and ensuring that public debate addresses them directly, creating pressure for solutions and accountability. The role of newspapers and the press becomes crucial in this environment, as they can help expose problematic content and inform the public about emerging threats. Candidates themselves also bear responsibility for discussing these limitations and considering solutions, both for the immediate election and for the next term. The stakes could not be higher: the 2026 elections will test not only the resilience of Brazilian democracy but also the ability of democratic institutions worldwide to adapt to a rapidly evolving technological landscape where the tools of manipulation have become democratized and the traditional safeguards of electoral integrity are being fundamentally challenged. As the campaign unfolds, the tension between technological capability and regulatory response will define the character of Brazilian democracy and potentially serve as a model for other nations grappling with similar challenges in an era of artificial intelligence and digital disinformation.
