When AI Meets Political Apathy: The Information Crisis We May Not See Coming

AI is becoming better at creating information. The bigger question is whether society will remain good at questioning it. Artificial intelligence is becoming easier to access. A student can use it to explain a difficult engineering concept. A business owner can generate marketing material in seconds. A developer can use it to write or debug code. A researcher can process enormous amounts of information with tools that would have seemed extraordinary only a few years ago. But AI is not only changing how people create information. It is changing how easily information can be manufactured. A photograph can be generated. A voice can be cloned. A video can be manipulated. A political statement can be fabricated. And increasingly, these things can be produced without requiring the technical expertise that was once necessary. That creates a problem that extends beyond technology. It creates a problem for societies whose citizens must make decisions based on information. The issue becomes even more important when people become politically disengaged and stop paying attention to the institutions around them. Because when people stop checking what is happening, somebody else gets to define what they believe is happening. This problem did not begin with AI Political manipulation of information is not new. Governments, political movements and other actors have used propaganda, selective information and misleading narratives for generations. Radio and television dramatically expanded the reach of political communication in the twentieth century. The internet then reduced the cost of distributing information globally. Social media went further by allowing almost anyone to become a publisher. AI represents another step in that progression. The important difference is not that humans suddenly discovered deception. It is that AI can dramatically reduce the cost, speed and technical barrier involved in producing convincing synthetic content. That distinction matters. A person once needed sophisticated editing skills to create a convincing manipulated video. Today, increasingly capable AI tools can do much of the technical work. The technology is therefore not inventing political manipulation. It is potentially industrialising it. We already have evidence This is no longer a purely hypothetical concern. The International Panel on the Information Environment analysed 215 documented generative-AI incidents across 50 countries that held competitive national elections in 2024. It found that 80% of those countries experienced reported GenAI incidents during their election periods. About 90% of the incidents involved content creation, including audio, images, videos and social-media posts. More than two-thirds were described as having a harmful role in the election. The researchers also found that 46% of incidents had no known source, while political candidates or parties were identified as sources in 25% and foreign actors in 20%. These figures should not be interpreted as proof that AI determined the outcomes of those elections. They show something different and arguably more important: AI-generated political content has already entered the information environment at significant scale. Deepfakes are becoming part of election campaigns The German Marshall Fund has separately tracked political deepfakes across elections around the world. Its database recorded 133 deepfakes across 30 countries and five continents, with 39% of elections in its tracking period featuring a significant deepfake campaign. Audio was particularly prominent, appearing in 69% of the tracked cases. That last figure deserves attention. People often imagine a deepfake as a sophisticated video in which a politician appears to say something they never said. But sometimes a convincing voice recording may be enough. A message forwarded into a WhatsApp group can be listened to in seconds. A person may not stop to investigate its origin. Someone else may forward it again. Then someone else. The technology does not need to convince everyone. It only needs to move faster than verification. One fake recording can create a real reaction Consider a hypothetical situation. An audio recording appears online showing a political candidate allegedly discussing plans to manipulate an election. The recording sounds authentic. It is shared thousands of times. People begin discussing it. Some demand an investigation. Others defend the candidate. News organisations begin checking it. Then, several hours later, experts determine that the recording was AI-generated. The recording was fake. But the reaction to it was real. That distinction is crucial. A false piece of information can produce genuine consequences even after the truth eventually emerges. And the problem becomes more difficult when citizens no longer know which sources deserve their trust. Nigeria has already seen the warning signs Nigeria does not have to look very far for examples of AI-assisted political misinformation. During the 2023 election period, researchers documented AI-generated political content in the Nigerian information environment. Freedom House reported an AI-generated audio recording that falsely purported to implicate an opposition candidate in election-rigging plans. The lesson is not that AI caused Nigeria's electoral controversies. It didn't. The lesson is that synthetic political content was already capable of entering Nigeria's political conversation before the technology became as widely accessible as it is today. The technology has continued to advance since then. 2027 will take place in a different information environment Nigeria is now approaching another general election. INEC has announced the timetable for the 2027 General Election, with the Presidential and National Assembly elections scheduled for January 16, 2027, and the Governorship and State House of Assembly elections scheduled for February 6, 2027. That makes the AI-information question particularly relevant. INEC itself has warned that elections are increasingly contested not only at polling units but also in the information space. In February 2026, the Commission specifically highlighted misinformation, disinformation and foreign information manipulation as threats to electoral processes. It also said that AI-driven information manipulation is part of the emerging challenge. And this is not merely an abstract concern. In April 2026, INEC warned Nigerians about fake social-media accounts impersonating its chairman and described forged digital interactions as part of a broader misinformation and identity-theft problem. The message is straightforward: Not everything carrying a familiar name, face or voice online should be assumed to be genuine. The more dangerous problem may not be believing everything There is another possibility that receives less attention. What if people stop believing anything? Imagine an information environment containing: A genuine government announcement. A fabricated government announcement. A real photograph. An AI-generated photograph. A genuine recording. A cloned voice. A legitimate election result. A fabricated election result. A correction to a false claim. A counter-claim saying the correction is itself false. At some point, some people may simply decide that nothing can be trusted. That creates what could be called an information exhaustion problem. People do not necessarily become committed to falsehood. They become tired of trying to determine the truth. And that can be just as consequential. Political apathy changes the equation Political apathy is not simply refusing to support one candidate or another. A citizen can disagree with every political party and still pay attention to what government institutions are doing. The more consequential form of disengagement is when people stop monitoring public affairs altogether. They stop checking official information. They stop following legislation. They stop investigating claims. They stop understanding how elections work. They stop asking who has the authority to make particular decisions. That creates an information vacuum. And information vacuums rarely remain empty. Someone eventually fills them. AI could make filling them easier. The attention problem This is where the AI skills gap becomes larger than a technology problem. AI literacy is often discussed in terms of prompting. How do you get better answers from an AI system? How do you automate a task? How do you generate images? How do you write code? Those are useful skills. But there is another form of AI literacy that may become increasingly important: Knowing when not to believe what you see. That requires source verification. It requires understanding context. It requires comparing independent sources. It requires recognising manipulated media. And sometimes, it simply requires being interested enough to investigate. The statistics tell only part of the story Numbers can show how widespread a phenomenon is. They cannot automatically tell us how much damage an individual piece of content caused. That distinction is important. The IPIE database records incidents; it does not establish that each incident changed an election result. The German Marshall Fund similarly notes that some deepfakes were not particularly convincing and that the impact of individual examples varies. This matters because the discussion should not become: "AI will determine elections." The evidence does not justify such a sweeping conclusion. A more defensible conclusion is: AI is increasing the volume, speed and sophistication of material that citizens and institutions may have to evaluate. That alone is a significant challenge. The problem is bigger than deepfakes Focusing exclusively on deepfake videos may actually cause us to miss the bigger issue. AI can generate ordinary text. It can create images. It can translate messages. It can imitate writing styles. It can produce large numbers of variations of the same message. It can assist with highly targeted communication. Some of these applications may be legitimate. Others can be deceptive. And not every misleading message needs to be technically sophisticated. Sometimes an ordinary photograph accompanied by a false caption can travel further than an expensive deepfake. The technology therefore needs to be considered as part of a much larger information ecosystem. What happens when the correction arrives too late? Suppose a fabricated claim spreads to 500,000 people in a few hours. A fact-checker later demonstrates that it is false. The correction is accurate. But perhaps only 50,000 people see it. The original claim has already reached ten times as many people. This illustrates a structural problem with information: Producing a false claim can be easier than disproving it. AI could widen that imbalance because generating content can become considerably cheaper than investigating it. The solution cannot simply be "fact-check everything." There is too much information. The public therefore needs a stronger first line of defence: people capable of questioning information before accepting or sharing it. The citizen becomes part of the verification system This does not mean every citizen needs to become a cybersecurity expert. It can begin with basic habits. Who published this? Where did it originate? Is there an official source? Are reputable independent outlets reporting the same thing? Is the audio or video being presented without context? Does the claim contain an extraordinary allegation without evidence? Has the supposed speaker actually published or said anything through a verified channel? And perhaps the most important question: Why am I being encouraged to react immediately? Urgency is powerful. So is outrage. So is fear. AI can potentially make content more convincing, but humans still provide the emotional response that gives information its social power. AI literacy should therefore include civic literacy The traditional definition of digital literacy is no longer enough. Someone can know how to operate a smartphone and still be highly vulnerable to manipulated information. Someone can know how to use AI and still be unable to distinguish an authentic political statement from a fabricated one. A stronger model would combine three forms of literacy. AI literacy: Understanding what artificial intelligence can generate and manipulate. Media literacy: Understanding how information is sourced, edited, distributed and verified. Civic literacy: Understanding how public institutions, elections and government decisions actually work. Together, these skills create something more valuable than technical knowledge alone. They create information resilience. Young Nigerians have a particular role to play Nigeria's large youth population means young people will occupy an important position in the country's digital information environment. They are among the most active users of social platforms and emerging technologies. That creates both opportunity and responsibility. A young Nigerian who understands AI can build useful technology. That same understanding can also make it easier to recognise when technology is being used deceptively. The objective should not be to become suspicious of everything. It should be to become curious enough to verify important things. The future may belong to people who can slow down The internet rewards speed. Social media rewards engagement. Algorithms often reward content that generates reactions. AI can make the production of such content dramatically faster. But truth does not necessarily move at the same speed. Sometimes the most intelligent response to an alarming piece of information is not to share it. It is to pause. Check. Compare. Then decide. That may sound simple. But in an environment where artificial intelligence can generate convincing content almost instantly, the ability to slow down could become a surprisingly valuable technological skill. And then there is 2027 Nigeria's next general election will take place in an information environment very different from the one that existed during previous election cycles. AI systems are more capable. Synthetic media is easier to produce. Social platforms remain major channels of political communication. And institutions are increasingly warning about information manipulation. None of this tells a citizen which political choice to make. It does raise a different question: How much attention should a citizen pay to the information environment in which those choices are being made? That is a question each person ultimately has to answer. Technology cannot replace civic judgment AI can produce information. It can analyse information. It can manipulate information. It can even help people identify manipulated information. But it cannot decide what citizens should believe. That responsibility remains human. A society that understands AI but pays little attention to public affairs could become vulnerable to people who understand how to exploit that technology. A society that combines technological knowledge with careful verification, media literacy and civic awareness has more tools for recognising manipulation when it appears. The biggest challenge may therefore not be stopping AI from creating information. That may be impossible. The bigger challenge may be ensuring that people remain capable of questioning the information AI helps create. Because when almost anything can be manufactured, perhaps the most valuable civic skill is no longer simply having access to information. It is caring enough to ask: Is this actually true?

Other insights from Galton

Contributor Points

Points-to-Coupons

Referral Earning