# The Anatomy of Election Misinformation: Lessons from Osun State
In the lead-up to Nigeria’s Osun State governorship election, something significant was happening beneath the surface of political campaigning. The information landscape wasn’t just noisy—it was being deliberately engineered to confuse, mislead, and manipulate voters. What unfolded over those weeks offers a disturbing preview of what Nigerian elections will increasingly look like, and it challenges our assumptions about how misinformation actually works in the digital age. The old model of fabrication—simply inventing stories out of thin air—was giving way to something far more insidious: the manipulation of real content, real images, and real events, repackaged to tell false stories.
The pre-election period was dominated by what fact-checkers call “context manipulation.” Instead of creating entirely fake photographs or statements, purveyors of misinformation took genuine material and stripped it of its true meaning. Take the claim that prominent APC chieftain Akin Ogunbiyi had endorsed the PDP candidate, Ademola Adeleke. The photograph circulating online was real—but it was from the 2022 election, not a current endorsement. Similarly, when a graphic purportedly showed candidate scores from an Arise TV town hall, it looked authentic enough to fool many viewers. But the graphic hadn’t originated from Arise News or the event’s organizers, and it conveniently excluded other candidates who had participated. This was misinformation with a sophisticated twist: it wasn’t about inventing events that never happened, but about twisting real events until they served a false narrative.
This distinction matters immensely for how we think about election integrity. When fact-checkers encounter a completely fabricated image, their job is relatively straightforward—they can prove it’s fake. But when an image is genuine but presented with false context, the verification process becomes much more complex. The question shifts from “Is this real?” to “When was this taken?” “Where?” “What was actually happening at that moment?” “Who originally published this?” “What is this actually evidence of?” These are harder questions to answer quickly, especially during the chaotic period of an election campaign when misinformation spreads at the speed of a WhatsApp forward.
## Election Day: When Speed Becomes the Enemy of Truth
As polling began, the nature of the problem shifted. The battleground moved from context manipulation to premature declaration. While votes were still being counted and no official results had been announced, viral claims began circulating about who was winning. One claim had Adeleke leading in 13 local government areas; another showed Bola Oyebamiji, the APC candidate, ahead. Both could not be true—and neither was verifiable at the time. But that was precisely the point. The goal wasn’t to present an accurate picture of the election’s progress; it was to shape public perception while the outcome was still genuinely uncertain.
The impact of this goes beyond mere confusion. When people see claims of victory before official results are released, it creates expectations that can be difficult to reverse. If supporters believe their candidate is winning and the official result later contradicts that belief, the gap between perception and reality becomes fertile ground for accusations of rigging and manipulation. Even if the official result is accurate and transparent, the earlier misinformation has already planted seeds of doubt. The result becomes suspect not because there’s evidence of wrongdoing, but because people’s expectations have been manipulated.
A particularly instructive example emerged from Modakeke, where a video circulated showing what appeared to be security intervention and the alleged removal of a ballot box. The video was real—genuine footage of an actual incident. But the claims attached to it went far beyond what the footage actually showed. The allegation that voters were being driven away by tear gas, or that a ballot box had been taken, could not be independently verified. The appropriate verdict was “partly true”—something happened, but not necessarily what the accompanying claims suggested. This case illustrates the spectrum of misinformation: it’s rarely all or nothing. The most effective falsehoods are usually built on a foundation of truth.
## The Post-Election Battle for Legitimacy
Once the election was over and Adeleke was declared the winner, the information war entered a new phase. The focus shifted from shaping the outcome to undermining its legitimacy. One early post-election claim alleged that the Returning Officer, Professor Joshua Ogunwole, had been arrested and detained in Abuja after declaring Adeleke the winner. This was entirely false—Ogunwole had traveled to Abuja, but there was no evidence of any detention. His university confirmed he remained free and at work. The claim appears to have been an attempt to delegitimize the result by suggesting that those who declared it were being punished.
Similarly, a photograph claiming to show APC governors holding an emergency meeting in response to their party’s defeat was exposed as an old image from a July meeting in Kebbi, unrelated to the election. And when results emerged showing that former Governor Olagunsoye Oyinlola had lost his polling unit, the information was accurate but was used to support a broader narrative that the entire election represented a wholesale rejection of certain political figures. The facts themselves were true; the interpretation attached to them was misleading.
Perhaps the most revealing post-election case involved Alhaja Falilat Yusuf, known as Ero-Arike, a prominent businesswoman and political figure. False figures circulated claiming that Accord had won her polling location with 241 votes to APC’s 138. The actual verified result was dramatically different—APC had won 392 votes to Accord’s 54. But what made this case particularly troubling was the gendered dimension of the attacks that followed. The false figures became ammunition for personal ridicule that went beyond political criticism, descending into sexist and humiliating language targeting her as a woman. This was misinformation doing double duty: misrepresenting an electoral fact while legitimizing personal attacks through a woman’s political standing.
## AI: The New Frontier of Political Manipulation
The most alarming development came in the form of a fabricated image showing Adeleke presenting his Certificate of Return to Peter Obi, the Labour Party presidential candidate. The image was AI-generated—a synthetic creation designed to imply a political relationship and endorsement that never existed. What was real was that Obi had congratulated Adeleke after the election. What was completely fabricated was the photograph and the story it told.
This is where AI represents a genuine step-change in political communication. The technology doesn’t just allow for the creation of false events; it allows for the manufacturing of false relationships. Who met whom, who endorsed whom, who helped whom win, who celebrated with whom—all these symbolic political moments can now be artificially created. In a political culture where imagery carries enormous weight, a convincing photograph can become evidence of something that never happened. The damage isn’t just in the specific falsehood but in the broader erosion of trust in visual evidence itself.
A subsequent fake scheme involving a supposed ₦15,000 cash transfer from Adeleke to celebrate his victory showed that the same technology could be weaponized for phishing. The link was not only false—it was designed to harvest personal information from unsuspecting users and encourage them to forward it to WhatsApp groups. The domain had been registered only days before the election. The information war was now an economic and cyber-security problem as well as a political one.
## The Real Weapon Is Doubt
Perhaps the most profound lesson from Osun is that AI’s greatest electoral power may not be its ability to make people believe things that are completely false. It may be its ability to make people doubt things that are true. This is the emerging “liar’s dividend”: once citizens know that politicians can be deepfaked, any genuine evidence can be dismissed as AI-generated. A real recording can be called fake. A genuine photograph can be described as manipulated. A legitimate result can be labeled fabricated.
This creates a corrosive information environment where the question is no longer simply “What is true?” but “Can we even agree on what counts as evidence?” When seeing is no longer believing, everything becomes suspect. It becomes possible to reject any unwelcome fact by invoking the possibility of AI manipulation. The very technology that enables more convincing lies also provides cover for dismissing inconvenient truths.
One of the most challenging aspects of this new landscape is the role of closed platforms like WhatsApp. While public misinformation on platforms like X (formerly Twitter) can be tracked and debunked in real-time, the most consequential misinformation may be circulating through private WhatsApp groups—ward groups, church groups, mosque groups, family groups, community groups. FactCheckAfrica itself acknowledged that monitoring these closed environments remained a major operational gap. The misinformation that spreads through private channels is the most difficult to counter because fact-checkers often can’t see it until it has already done its damage.
## Preparing for 2027: Beyond Better AI Detectors
The Osun experience should serve as a warning for Nigeria’s 2027 elections. The information battle is becoming faster, more personalized, and harder to monitor. But the response cannot simply be better AI detection technology. Detection will always be playing catch-up with production. Every detector can be defeated by a better generator.
The more sustainable approach is to strengthen the entire information ecosystem. This means investing in local journalists who understand verification techniques. It means training election observers to provide ground truth that can counter online falsehoods. It means building institutions that can communicate quickly and authoritatively. It means developing fact-checking capacity with forensic sophistication. It means pressuring platforms to respond to coordinated manipulation. And it means educating citizens to understand that a viral post is a claim, not evidence.
The TruthGuard model demonstrated the value of connecting these various layers. Field observers could provide context for what was actually happening on the ground, while fact-checkers tested what was being claimed online. Crucially, citizens could feed suspicious material directly into the verification process, creating a participatory system of accountability. FactCheckAfrica took this further by converting its checks into short graphics and distributing them through a network of more than 70 trained journalists, ensuring that corrections reached the same networks that had carried the falsehoods.
But there are limits to what fact-checking can achieve once a lie has traveled. The correction must reach the same networks that carried the original falsehood—and ideally, it must arrive quickly enough to prevent the lie from taking root. This is why the most important investment may be in building a culture where citizens pause before forwarding, journalists verify before amplifying, political actors resist the temptation to weaponize falsehood, institutions respond rapidly, and technology companies recognize that information manipulation is not a side issue but a core part of electoral infrastructure.
Osun showed us something essential about the nature of modern election manipulation. Misinformation doesn’t have to change every vote to damage democracy. It can work its corrosive effect by creating fear before people vote, confusion while they vote, and doubt after the votes have been counted. That may be the most sophisticated form of AI-enabled political manipulation yet—not the dramatic fake video that everyone notices, but the slow, steady erosion of confidence in the democratic process itself. When citizens can no longer trust what they see, hear, or read, the very foundation of informed consent begins to crumble. And that is a threat that no fact-checker alone can counter.