BY: Mustapha Lawal
In the weeks leading to Nigeria’s 2023 general elections, social media users encountered an endless stream of political videos, manipulated images, and sensational claims about candidates. Some were authentic. Others were edited. Many were completely fabricated.
Today, the challenge has evolved. Artificial intelligence has made it possible to generate realistic images, videos, voices, and even entire personalities that never existed. Unlike traditional misinformation, these new forms of synthetic content are often not designed to deceive people into believing they are real. Instead, they are designed to make people feel that they are true. And that distinction may be more dangerous.
Researchers studying the rise of political deepfakes have found that many consumers engage with AI-generated content not because they believe it is authentic, but because it reinforces what they already think, fear, or hope is true. As Nigeria’s digital ecosystem expands and political conversations increasingly move online, understanding how deepfakes influence public opinion has become a critical media literacy challenge.
The New Face of Misinformation
For years, misinformation in Nigeria largely revolved around manipulated photographs, recycled videos, fabricated quotes, and false claims. Fact-checkers have repeatedly debunked old videos falsely linked to elections, foreign conflict footage presented as attacks in Nigeria, and fabricated statements attributed to politicians and public officials.
Artificial intelligence is changing that landscape. Today, a person with little technical expertise can generate a convincing image of a politician attending an event that never happened, deliver a speech they never made, or appear in a setting they never visited.
The technology has become increasingly accessible, faster, and cheaper. Researchers at Purdue University’s Governance and Responsible AI Lab (GRAIL) recently documented a dramatic rise in political deepfakes. Since the beginning of 2025 alone, the lab has recorded more than 1,000 English-language social media posts containing AI-generated political imagery and videos. In the previous eight years combined, researchers identified just over 1,300 similar incidents. The trend points to a fundamental shift in how information is created and consumed.
Nigeria Is Already Witnessing the Effects
While Nigeria has not yet witnessed deepfake campaigns on the scale seen in some Western elections, early warning signs are already visible. During election periods, images purportedly showing candidates in secret meetings frequently emerge online. Fabricated endorsements appear on social media. Edited videos are circulated to suggest politicians made controversial statements.
Nigeria has already witnessed how convincing political deepfakes can become when they align with existing beliefs.
During the 2023 presidential election, AI-generated videos and images depicting global celebrities holding placards in support of Labour Party candidate Peter Obi circulated widely across social media. Although the content was fabricated, many supporters accepted it as authentic and shared it extensively. In a street experiment conducted by Zikoko Citizen, members of the public were shown some of these AI-generated visuals and asked whether they believed they were real. Several respondents not only accepted the content as genuine but also indicated they would share it with others because it reinforced their conviction that Obi was the most qualified candidate.
Another similar experiment involving AI-generated videos purporting to show military deployments against protesters produced comparable results, with some viewers failing to identify the synthetic nature of the footage. These examples illustrate a growing challenge in Nigeria’s information ecosystem: people often evaluate content not based on whether it is authentic, but on whether it confirms what they already believe to be true.
Nigeria’s insecurity crisis has created fertile ground for synthetic content. FactCheckAfrica investigations have documented numerous instances where videos from Burkina Faso, Sudan, Niger Republic, and other conflict zones were falsely presented as evidence of attacks in Nigeria.
As AI tools become more sophisticated, distinguishing between genuine footage and fabricated content will become increasingly difficult. The danger is not only that people may believe false information. The greater risk is that people may accept fabricated content because it aligns with existing beliefs.
Why Deepfakes Work Even When We Know They Are Fake
This tendency is consistent with what researchers describe as “belief reinforcement.” Valerie Wirtschafter of the Brookings Institution argues that deepfakes do not necessarily change minds; rather, they strengthen pre-existing opinions and make people less likely to question information that supports their worldview.
Daniel Schiff, an assistant professor of technology policy at Purdue University, argues that AI-generated content often succeeds because it creates narratives that audiences already find emotionally plausible.
“A lot of people feel like these images or videos or the stories they convey feel true,” he observed.
This psychological dynamic helps explain why deepfakes can be influential even after viewers learn they are fake. People do not always consume information as investigators searching for objective truth. More often, they consume information through the lens of identity, ideology, religion, ethnicity, political affiliation, and personal experience.
If a manipulated video appears to confirm what someone already believes about a politician, government institution, religious group, or ethnic community, they may continue sharing it even after its authenticity has been challenged.
In other words, the emotional impact often survives the factual correction.
The Danger of Artificial Consensus
Researchers describe one emerging threat as “AI swarms.” These are networks of AI-generated accounts capable of posting content, engaging users, amplifying narratives, and reinforcing each other’s messages without direct human involvement.
Valerie Wirtschafter of the Brookings Institution compares the phenomenon to traditional troll farms but without the need for large numbers of human operators.
“It is sort of like a troll farm without actually having to have people anymore,” she noted.
For countries like Nigeria, where misinformation has previously fueled electoral tensions, religious conflicts, and ethnic divisions, such developments could have serious consequences.
Artificial consensus can create the impression that certain viewpoints are universally accepted when they are not. Repeated exposure to those viewpoints may then influence how citizens perceive public opinion itself.
The timing is particularly important. As Nigeria approaches electoral cycles, AI-generated political content is likely to become increasingly common. The challenge is compounded by the speed of digital communication. False content often spreads faster than corrections. By the time verification occurs, the narrative may already have reached thousands or even millions of people.
The Technology Is Not the Problem Alone: How Nigerians Can Protect Themselves
While much attention focuses on AI itself, experts argue that technology is only part of the challenge. The deeper issue is human behaviour. People often share information because it provokes anger, validates beliefs, or aligns with group identities.
In such circumstances, whether content is authentic sometimes becomes secondary. This is why media literacy remains essential. Technology can generate deepfakes. But people decide whether to believe them.
As synthetic media becomes more common, citizens must develop new habits for evaluating information. Before sharing political content, ask:
- Who originally posted it?
- Is the source identifiable?
- Can the claim be independently verified?
- Are official sources confirming it?
- Has a reputable media organisation reported it?
- Does the video show signs of manipulation?
Most importantly, resist the pressure to share immediately. Verification often requires time. Virality does not equal authenticity.
The Future of Truth in the AI Era
The rise of deepfakes does not mean society is doomed to misinformation. Researchers continue to develop authentication technologies capable of tracing the origins of digital content. Fact-checkers are expanding their use of AI detection tools. Platforms increasingly require labels for synthetic media.
But none of these solutions will be sufficient on their own. The ultimate defence against misinformation remains an informed and skeptical public. As AI blurs the line between reality and fabrication, citizens must learn a new digital skill: distinguishing between what is real and what merely feels true. In an information environment shaped by algorithms, synthetic media, and emotional persuasion, that ability may become one of the most important forms of civic literacy in the twenty-first century.
Editor Note:
This article is part of FactCheckAfrica series examining the growing harms of artificial intelligence (AI) across Africa, with a particular focus on Nigeria. The series is exploring how AI is amplifying existing inequalities, distorting information ecosystems, and reshaping power in fragile democracies.
