Picture a man in Sri Lanka, sitting at a laptop, learning what outrages British people. He doesn’t live there. He doesn’t speak English fluently. He has never walked past a Victorian terrace or stood in a fish and chip queue. But he knows, with algorithmic precision, what will make someone in Manchester feel that their country is being taken from them.
He feeds that knowledge into an AI image generator. A sepia-tinted image appears: Victorian London, the streets narrow and cobbled, the caption mourning a time when the city “was English, first-world and beautiful.” He posts it to a Facebook page called something like Britain Today. The union jack is the profile picture. He waits. The engagement comes. The Stranger Who Studied You First.
According to a months-long investigation published in The Guardian by Niamh McIntyre of the Bureau of Investigative Journalism, there are hundreds, possibly thousands, of these Facebook pages targeting British audiences. Behind many of them are young entrepreneurs from Pakistan and Sri Lanka, using generative AI tools to manufacture reactionary British content at scale. One creator, Geeth Sooriyapura, claimed to have made $300,000 over his Facebook career. Another, a Pakistani creator identified only by his page’s activity, reportedly earns $1,500 a month from a single page. Both were making many times the average income in their home countries. Meta’s ad revenue sharing and engagement bonuses made it all possible.
The detail that keeps surfacing, though, is not the money. It’s this: one of these creators is a devout Muslim. He produces content calling Islam a “cancer.” He does it because the algorithm rewards it.
What do you do with that?
There’s something almost philosophical happening here, something that cuts well past “foreign disinformation” as a category. We’ve spent years watching algorithms learn what we want, then slowly bend us toward wanting more of it. The usual critique is that recommendation engines trap us inside our own preferences, reflecting distorted versions of ourselves back at us until we can’t tell the difference between what we believed and what we were led to believe. That critique assumes there’s a real you somewhere in the system, a baseline self being slowly warped.
This story suggests the loop is more severed than that. The reflection has been outsourced entirely. The person holding the mirror doesn’t just not care about the image it shows. They actively don’t share any of the experiences, fears, or identities the image is supposed to represent. They have studied the emotional signature of British nationalist anxiety with the clinical detachment of a market researcher. They know what a union jack does to a certain type of Facebook user. They know that sepia makes nostalgia feel ancient, and ancient makes it feel real. They know the phrase “stopping serving pork” will light up more comments than almost anything else. The Mirror That Learns.
The algorithm taught them this. And the algorithm made it worth their time to act on it.
What’s strange to sit with is how effective it remains despite that gap. Comments beneath these videos call for Muslims to be deported, fantasize about ethnic civil war. People watching AI-generated content produced by a Pakistani Muslim for profit are experiencing what feels to them like community, validation, recognition. They feel seen by something that was never looking at them at all. The content doesn’t need to come from shared feeling to produce shared feeling. It just needs to fit the shape of the feeling that already exists.
That’s the part that should unsettle anyone who thinks carefully about identity and technology. We have a working model now of how to manufacture the sense of belonging to a group without having any relationship to that group. You study what makes the group feel cohesive and threatened, you replicate the emotional cues with AI precision, and you publish it somewhere their algorithm will surface it. If the emotional signature is accurate enough, the felt experience is indistinguishable from genuine solidarity.
This is what Sooriyapura was teaching 2,500 students in his content academy. He described how AI-generated political videos can go viral up to ten times faster than other content. What he was really teaching, whether he framed it this way or not, is a theory of identity manipulation. He was teaching his students how to locate the shape of someone’s fear and fill it with generated content until it solidifies into belief.
The infrastructure that made this scalable has two parts. First, generative AI brought the production cost of convincing-looking cultural content close to zero. You no longer need to understand the culture to produce artifacts that look like they came from inside it. Second, Meta’s retreat from content moderation removed the friction that might have slowed it down. Trust and safety teams were cut. The platforms pulled back. The space that opened didn’t stay empty.
What keeps surfacing is the “passive income” framing that shapes this entire industry. McIntyre notes that “passive income” culture runs through these operations: the promise that you can quit your job and make easy money online, that success is a matter of finding the right system. The creators selling courses on how to produce algorithmic ragebait are selling a process, not a belief. There’s nothing ideological about it, from their perspective. Britain’s fear of demographic change is a market segment. So is American evangelical anxiety. So is French nostalgia for some more coherent national past. Identify the emotional niche, generate the content, collect the revenue.
This is the endpoint of treating attention as a pure resource. When engagement is the metric that drives payment, and when rage drives more engagement than any other emotion, the rational move for anyone trying to monetize attention is to study what people hate and give them more of it. The devout Muslim posting Islamophobic content for $1,500 a month is not a hypocrite. He’s a rational actor inside an irrational system. The irrationality is structural, not personal.
The harder question, the one without a clean answer, is what this does to the people consuming the content. They are forming something they experience as identity, as belonging, as grievance with a coherent shape. That experience is real even if everything that produced it was manufactured. The anger is genuinely felt. The fear is genuinely felt. The sense of having one’s concerns validated is genuinely felt. None of it was sparked by anyone who shares those concerns.
We tend to think of radicalization as something that happens to vulnerable people through sustained exposure to extremist communities. This is different. It’s systematic delivery of the emotional raw material of radicalization, at scale, through an advertising network, by people who have mapped your fear from the outside and learned to speak its language without ever having lived inside it.
The question that remains is not how to stop them. The question is what it means that the emotional architecture of your political identity can be accurately modeled, mass-produced, and delivered to you by someone who couldn’t pick you out of a crowd. What part of you, exactly, is being reflected when the mirror is held by a stranger?
Digital Alma explores technology, consciousness, and what it means to be human in a digital world.
Leave a Reply