The modern information landscape has become a strange and unsettling theater, where the lines between genuine human connection and manufactured digital performance are blurring beyond recognition. At the heart of this transformation is a new kind of operator: the digital manipulator, a person who understands that the most powerful deception is not a single fake post but an entire ecosystem of fake interactions designed to look like a living, breathing community. Creating this sense of organic engagement is now a central part of the manipulator’s craft, and it has been made dramatically easier by the rapid advancement of artificial intelligence tools. Consider the words of Adeyemi Abduwahab, a Lagos-based influencer who openly describes the strange, almost lonely mechanics of his work. He paints a picture of a carefully choreographed cascade: one bot posts a provocative statement, another immediately supports it, a third jumps in with a counter-argument, and a fourth swoops in to fact-check the whole exchange. To an outsider scrolling through their feed, it looks like a vibrant, spontaneous debate among passionate and independent users. In reality, as Abduwahab puts it, “it’s just me playing chess with myself.” This confession reveals the deeply artificial nature of online discourse today, where the appearance of disagreement can be manufactured to create excitement, tension, and ultimately, engagement that can be directed toward any goal. The challenge, however, is that raw automation is too clean, too predictable, and platforms have become increasingly sophisticated at spotting such robotic behavior. This is where the human touch becomes indispensable, and it is a delicate, almost artistic process. The manipulator must deliberately program imperfection into the bots: strange delays between messages, accidental-looking typos, an overuse or misuse of emojis, the casual slang of a particular region, and even the rambling, slightly incoherent posts that people typically write late at night when they are tired or emotional. These small flaws are what make the fake accounts feel real, and they are crucial for deceiving both the platform algorithms and the human users who will eventually encounter the content.
The effort to make robots act human goes far beyond just the text they produce, extending into the very technical and behavioral fingerprints that define an online identity. Platforms like Facebook have made coordinated inauthentic behavior a punishable offense, removing fake accounts and networks that are found to be operating in a deceptive manner. To evade these detection systems, manipulators have had to develop a kind of digital camouflage that is both intricate and obsessive. As Abduwahab explained, every single account in his network has its own unique fingerprint, a set of characteristics that makes it distinguishable from every other account, just as each person has a unique physical appearance. This involves using a distinct browser profile, rotating IP addresses so that no single connection seems suspicious, and even simulating false sleep patterns, so that an account goes quiet at odd hours and then resumes activity in a way that mimics a human’s daily routine. The bots do not simply post content into the void; they are trained to engage in all the small, mundane rituals of social media life. They comment on other people’s posts, they like the content of their supposed friends, they wish each other happy birthday, and they generally participate in the social graces that most of us take for granted. It is an elaborate performance designed to build a kind of artificial credibility, a history of normal behavior that makes the account seem trustworthy. When all these elements are combined, the result is a synthetic persona that can successfully insert itself into real conversations, influence opinions, spread rumors, or amplify a particular message without ever being detected. The sheer level of detail involved in this process is a testament to the deep arms race that is constantly being waged between the manipulators and the platforms tasked with protecting their users, a battle in which the manipulators seem to be constantly refining their tactics and moving one step ahead.
Language is perhaps the most powerful weapon in the digital manipulator’s arsenal, and artificial intelligence is making it easier than ever to use language in ways that feel deeply authentic and culturally specific. In a multilingual country like Nigeria, where vast populations communicate in languages such as Hausa, Yoruba, Igbo, and Nigerian Pidgin, the ability to speak these languages naturally is essential for making fake content feel real. Large language models, the same technology that powers modern AI chatbots, are being trained to understand and generate text in these languages, and a surprising part of this training is happening thanks to the very influence entrepreneurs who will later use these tools for deceptive purposes. Nigeria’s so-called influence entrepreneurs are actively feeding these AI models with thousands of realistic WhatsApp conversations, dialogues that are conducted in a mix of local languages and peppered with the ungrammatical, casual phrasing of everyday life. This is a brilliant and cynical move. By training the AI on real human conversations, they ensure that the bot’s output is not stilted or textbook-perfect, but instead sounds like a real person texting a friend. For example, when a bot replies with something like “omo, na wa o,” a phrase that blends Yoruba and Nigerian Pidgin to express a kind of overwhelmed amazement, it is nearly impossible for a human reader to distinguish that response from one written by a person sitting in Lagos, Kano, or Port Harcourt. This deep linguistic authenticity is a massive advantage for disinformation agents, because content that feels locally familiar is far less likely to be questioned. The same technology can be used to replicate the voice and mannerisms of prominent public figures, with just a few manual tweaks allowing a creator to produce audio that sounds incredibly close to a real politician or celebrity. One influential creator based in Abuja, Nigeria’s capital, claimed during an interview that AI now automates about seventy percent of his work, with humans handling the remaining thirty percent. He noted that this ratio is likely to shift even further toward AI as its capabilities continue to grow, which suggests that the barriers to entering this field are falling rapidly.
This democratization of disinformation means that the old stereotype of a sophisticated, secretive propaganda headquarters is completely outdated. The reality is far more mundane and, in a way, more frightening. A Kaduna-based content creator named Ibrahim, popularly known as Iromba, captured this new reality with a laugh, pointing out that people imagine secret rooms filled with teams of specialists, but often the entire disinformation operation is just one person, a single boy in Kaduna, aided by a handful of powerful AI tools. This is not to diminish the danger of what is being done, but rather to highlight that the scale and complexity of the threat have fundamentally changed. In the past, creating a large-scale disinformation campaign required significant resources, technical expertise, and perhaps the backing of a state actor or a wealthy organization. Today, it requires a smartphone, an internet connection, a few affordable AI subscriptions, and a decent understanding of the target audience’s culture and language. The result is that the information environment is becoming increasingly polluted, and the ability to spread falsehoods is being placed in the hands of people who may have no political agenda at all, only a desire for profit, influence, or perhaps just the thrill of manipulating others. This new breed of manipulation is not just about changing political outcomes or destabilizing communities; it is also a business. Fake engagement can be sold to advertisers, political candidates, and anyone else who wants to create the illusion of grassroots support. The combination of AI’s raw power and the human ability to add subtle, authentic touches is giving rise to a generation of digital entrepreneurs who can build entire audiences out of thin air, and the implications for democratic societies are profound.
As this technology races ahead, regulation and fact-checking efforts are falling dangerously behind. Platforms and fact-checkers are already playing catch-up, but the sheer speed and sophistication of these AI-assisted manipulation techniques are distancing them further with each passing day. One of the most concerning aspects is the difficulty of reversing a false claim once it has gone viral, particularly in private messaging apps like WhatsApp, where content spreads rapidly among trusted social groups. A single doctored picture or a cleverly manipulated audio clip can travel to hundreds of thousands of people in a matter of hours, and even if fact-checkers eventually prove it to be false and circulate that information widely, the damage has already been done. The falsehood has been seen, understood, and internalized by a massive audience, and a correction later simply cannot compete with the immediate emotional impact of the original fake. Platforms are also struggling to build systems that can detect falsehoods in the many languages used around the world. A fact-checker in Abuja, Bilkisu Sarina, noted how their systems sometimes catch English-language posts, but when the content is in Hausa, Yoruba, or Pidgin, the detection systems almost never flag it. This is the exact environment where disinformation thrives, because the tools that should be protecting users are effectively blind to the languages that those users speak. There is also a more cynical question about the motivation of the platforms themselves. Many of these social media companies operate on business models that reward virality and engagement above all else, meaning that a piece of content that sparks outrage, confusion, or excitement is considered good for business, regardless of its truthfulness. This creates a fundamental conflict of interest, where the mechanisms meant to monetize content are actively incentivizing the spread of misleading information.
Ultimately, this is not just a technological problem, but a deeply human one, rooted in our innate desire to connect, to trust, and to feel part of a shared story. The manipulators understand this all too well, and they exploit it by crafting their content to feel familiar, to tap into existing fears, prejudices, and desires, and to present themselves as members of our own community. The balance between AI automation and human effort is shifting, and as AI continues to improve, it will become increasingly difficult for ordinary users to protect themselves through sheer skepticism alone. We may reach a point where it is impossible to know whether the person arguing with you in the comments section is a human being or a ghost in the machine, built to steer the conversation in a particular direction. The answer cannot be found in better technology alone, although that will certainly help. It requires a new emphasis on media literacy, teaching people to question the sources of their information, to verify claims through multiple channels, and to be suspicious of content that seems designed to provoke an emotional reaction. It also requires pressure on the platforms to act responsibly, to prioritize accuracy over engagement, and to invest the resources needed to monitor and remove inauthentic behavior across all languages and cultures. The fight against this new wave of digital deception will be long and difficult, and perhaps it will never be fully won. But by understanding the techniques used by these digital manipulators, by recognizing the human flaws and desires they exploit, and by acknowledging that disinformation is no longer the work of shadowy elites but often just a single person with an AI tool, we can begin to build the resilience needed to navigate this strange and uncertain new world.