It is hard to believe that just a few years ago, the phrase “fake news” was enough to describe the digital chaos swirling around elections. Those days are gone. According to a major new Brazilian report, the threat has mutated into something far more slippery, more realistic, and more dangerous. The Brazilian Internet Steering Committee, known by its Portuguese acronym CGI.br, released its updated “Internet, Democracy, and Elections Guide” on September 22, and the conclusions are sobering. The report argues that fake videos created by artificial intelligence and low-quality information have replaced traditional fake news as the main digital threat to elections and democracy. In other words, we are no longer dealing with poorly written headlines or obvious hoaxes shared by careless relatives. We are dealing with synthetic media that looks and sounds real, combined with a flood of shallow, misleading content that is hard to trace and harder to debunk. The guide compares the current election cycle with the one in 2018, when the first edition was published, and the contrast is stark. Back then, open social networks like Facebook were the main channels for spreading false information. Today, the report says, more restricted and private networks like WhatsApp have become the primary vehicles for disinformation. This matters because private groups are invisible to fact-checkers, researchers, and the broader public, making it much easier for false narratives to circulate unchallenged. The bad news is that this is not a Brazilian problem alone; it is a global phenomenon, but Brazil has become a kind of laboratory for seeing how digital manipulation can shape political reality.
The guide goes deeper, showing that disinformation has become increasingly widespread and now moves simultaneously across multiple platforms. It no longer stays in one corner of the internet. A lie can start in a WhatsApp group, appear as a video on TikTok, get picked up by a Telegram channel, and then circle back to Facebook and Instagram, all within hours. In terms of format, the researchers noticed a clear trend: video has become the primary vehicle for disinformation, almost always accompanied by a short text that anchors and legitimizes the false narrative. That little caption matters. It gives the video context, makes it feel like evidence, and gives viewers a ready-made interpretation before they can think for themselves. This is a far cry from the meme-and-headline era of 2018. The report also notes that institutions like the World Economic Forum have already considered disinformation one of the greatest global risks for 2025. That is a warning, not just a headline. If the world’s economic and political leaders are placing disinformation on the same level as climate change and armed conflict, then we should all pay attention. The report’s authors are not just worried about the content itself; they are worried about what it does to the way people think and trust. When false information is everywhere, and when it is polished, emotionally gripping, and tailored to specific audiences, it becomes harder for ordinary citizens to separate fact from fiction. The central role of the internet in everyday life—how we get news, how we entertain ourselves, how we communicate with friends and strangers—makes this an even more pressing issue for democratic contests. The internet is not a separate virtual world anymore; it is the main arena where public opinion is formed, and that makes it a battleground for the very idea of democracy.
Recent technological developments have only intensified the danger. Tools such as generative artificial intelligence emerged during the 2022 election cycle, but the report expects them to have an even greater impact this year. Why? Because these tools are now being combined with highly sophisticated dissemination strategies. It is not just about creating a deepfake; it is about targeting that deepfake to the exact people who are most likely to believe it, and then wrapping it in a narrative structure that feels authentic and familiar. The report describes how generative AI can produce custom-made content for specific user profiles, making each piece of disinformation feel personally relevant. This is a huge shift from the old spray-and-pray approach, where the same false story was sent to everyone. Now, a person who cares about farming might receive a fake video about agricultural policy, while a person concerned about crime might receive another fake video about public safety, each crafted to exploit their particular fears and hopes. The result is a more fragmented public square, where different groups live in completely different realities. Renata Mielli, the coordinator of CGI.br, put it clearly when she said that digital platforms now play a central role in how Brazilians stay informed about the world, their country, and their own cities. But she also stressed the responsibility of private companies, which are often driven by economic objectives rather than the public interest. These platforms are not neutral pipes; they are businesses built on engagement, and engagement often rewards outrage, fear, and lies. Mielli argues that understanding how these environments function, how content is distributed, and what impact this has on public debate is essential to strengthening the integrity of information and democracy itself. That, she says, is one of the main goals of the guide: to clarify the hidden mechanics of the digital world and give citizens, journalists, and policymakers the tools to understand what is happening.
The guide itself is a practical resource, available on the CGI.br website, and it covers a wide range of topics that are essential for anyone trying to navigate the current information landscape. It explains how content recommendation systems operate, shedding light on the opaque algorithms that decide what we see and what we never see. It also examines the impact of generative artificial intelligence on the electoral landscape, showing how tools like deepfakes and conversational language models work under the hood. For example, the report breaks down how deepfakes are made, why they are so convincing, and why they are so difficult to detect, even for experts. It also looks at conversational language models, the kind of AI that can simulate human conversation, and how they can be used to create fake news, impersonate real people, or generate endless variations of a propaganda message. The guide does not stop at diagnosis. It also includes clear guidelines for institutional efforts to combat disinformation, suggesting what governments, courts, election authorities, and civil society organizations can do to push back. And it offers practical advice for ordinary citizens on how to avoid becoming victims of false or misleading content—and, just as importantly, how to avoid becoming spreaders of it. This is crucial because disinformation does not spread by itself; it spreads because people share it. Often, they share it not out of malice, but out of fear, anger, or a sincere belief that they are helping others by sharing an important warning. The guide also analyzes the structures of media outlets and social networks, highlighting how they differ in their responsibilities and their vulnerabilities. One of the most interesting sections focuses on the role of influencers, the new communicators of the digital age. The report acknowledges that influencers have a prominent role in shaping public opinion, but it points to a fundamental difference between them and traditional journalists: influencers do not follow a code of ethics, and they are not institutionally accountable for the accuracy of their statements. If a journalist publishes a lie, there are professional consequences; an influencer who publishes a lie can simply move on, delete the post, or double down. That asymmetry is dangerous, especially during elections.
The report’s warning about deepfakes deserves special attention. According to the CGI.br study, the most frequently cited concern regarding deepfakes is their realism or credibility. As this type of deceptive content has undergone constant refinement, it now has the potential to deceive even voters who are highly experienced in evaluating content. In the past, you could spot a fake by looking for blurry edges, unnatural movements, or weird lip-syncing. Today, AI-generated videos can be virtually indistinguishable from real recordings, especially when they are watched on a small phone screen and shared in a hurry. This is not just a threat to individual elections; it is a threat to the very idea of truth. When people cannot trust what they see with their own eyes, they begin to doubt everything, including legitimate journalism and official information. The guide warns that this type of content can foster widespread distrust of all kinds of information, including truthful information. This is perhaps the most insidious effect of deepfakes: they do not only spread lies, they also make it easier for people to dismiss the truth as just another fabrication. When a real video surfaces of a politician saying something damaging, the response is often, “It could be a deepfake.” That doubt is enough to neutralize the impact, even if the video is perfectly genuine. The report connects this loss of trust in institutions such as the press and the judiciary with political polarization. Together, these forces create a climate in which personal convictions take precedence over fact-checking. People stop asking “Is this true?” and start asking “Does this support what I already believe?” If it does, they share it; if it does not, they dismiss it. This is exactly the kind of environment where disinformation thrives.
Another problem that arises from this climate of mistrust is the way people increasingly consume content that confirms what they already believe. The report describes a phenomenon known as “motivated reasoning,” which is a fancy way of saying that people are biased when they evaluate new information. When people encounter information that contradicts their existing beliefs, they tend to reject it, often without really thinking about it. But when they encounter information that reinforces those beliefs, they accept it eagerly, even if it is false. In this sense, disinformation can reinforce and radicalize ideologies, making it harder and harder for people to change their minds or even to have a productive conversation with someone on the other side. The digital environment makes this worse because recommendation algorithms are designed to keep people engaged, and one of the best ways to keep someone engaged is to show them content that makes them feel right, angry, superior, or afraid. Over time, users are pushed into echo chambers where the same messages are repeated over and over, and the wider public conversation becomes fragmented and hostile. The CGI.br guide does not pretend to have easy answers, but it insists that we are not helpless. By understanding how these technologies work, how platforms distribute content, and how human psychology responds to deceptive media, we can build defenses. Institutions need to invest in fact-checking, media literacy, transparency, and accountability. Citizens need to slow down, question their own instincts, and remember that just because a video looks real does not mean it is real. Most of all, the report makes clear that democracy cannot survive, in any meaningful way, if the public cannot agree on basic facts. The fight over information is not a side issue; it is the central struggle of our time. The guide is a reminder that we all have a role to play, not only as voters, but as readers, viewers, sharers, and members of a common public space. In the end, the question is not whether artificial intelligence can create convincing lies—it clearly can. The question is whether we can build the social, legal, and educational tools to keep those lies from destroying the trust that makes democracy possible.