Your Algorithm Groomed Someone You Love Into a Cult. Here's the Mechanism.
Eight months. That's how long it took.
Your brother watched a few videos. Started commenting in forums. Made some new online friends. And eight months later he was unrecognizable. He stopped trusting the people closest to him. He had an answer for everything — and none of it sounded like anything he'd thought before. When you pushed back, he told you that you just didn't understand. That he finally did.
You couldn't figure out what happened to him.
Frontiers in Political Science published a 2026 framework documenting how extremist groups leverage AI recommendation algorithms to systematically target psychologically vulnerable people — following the same psychological architecture as cult indoctrination, step by documented step. The algorithm didn't radicalize your loved one. But the algorithm did what cults used to have to do with in-person recruiters: find the wound, offer the salve, and guide the person deeper.
Step One: Find the Wound
No algorithm is searching for people to radicalize. It's searching for engagement — specifically, people whose emotional response to content is strong enough to keep them watching and clicking.
Psychologically vulnerable people engage strongly. People who feel invisible, who feel wronged, who carry unnamed fear — they don't scroll past content that speaks to those feelings. They stop. They watch again. They share. The algorithm reads this as a signal: more of this. More content about unfairness. More content about who's responsible. More content that tells you that what you've been feeling has a name and a cause.
The recommendation engine has no ideology. It has a feedback loop, and the feedback loop learns your wound faster than most therapists would.
The content that finds your vulnerable person isn't usually extreme at the start. It's validating. It's finally someone saying what they've been thinking but couldn't put into words. That's the entry point — not the belief, but the relief of recognition.
Step Two: Offer the Community
After the content comes the community.
Comment sections, forums, group chats — the algorithm surfaces these as naturally as it surfaces the next video. And the community that forms around this content offers something that's genuinely powerful: belonging, recognition, and significance.
If you've ever felt like you didn't matter — like you were invisible, replaceable, overlooked — and someone said "we see you, you understand things others don't, you're one of the few who get it" — that's not nothing. That's one of the most powerful psychological needs a human has, being met. The fact that the recognition comes with conditions isn't obvious yet. The conditions come later.
Research on cult recruitment has documented this sequence for decades: you don't recruit with ideology. You recruit with belonging. The way apps exploit attachment anxiety to drive engagement follows the same principle — the community becomes the thing people are loyal to, and the content of that community becomes secondary to the belonging itself.
Step Three: Narrow the World
This is the part that looks like a personality change from the outside.
The algorithm narrows first. Every recommendation is more of the same, steeper in the same direction. The outside world — the news sources your loved one used to read, the friends they used to talk to, the perspectives that used to balance their thinking — starts to feel less real than the world inside the feed. Not because they consciously chose to close themselves off. Because the feed delivers a complete world with consistent internal logic, and the outside world delivers noise.
Then the community narrows. People in these spaces almost universally develop a framework for discrediting outside sources. It's not censorship or paranoia from the inside — it's discernment. Those people don't understand. Those institutions are captured. You have to look at who benefits. The framework is self-sealing: every counterargument becomes evidence of how captured the arguer is.
By the time your loved one stops trusting you, they're not being irrational. They're applying a consistent framework that the algorithm and the community built together, one video and one comment at a time.
Step Four: Install an Identity
The final step is the one that makes exit hardest.
Somewhere in this process, the ideology stopped being a set of beliefs and became an identity. "I'm the kind of person who sees through this." "I'm one of the people who understands what's really happening." When someone offers a counterargument, they're not challenging a belief — they're challenging who this person is.
This is precisely how cults work. The belief system isn't separate from the self; it's constitutive of the self. Leaving the belief means losing the identity, the community, and the significance that came with it. No argument is strong enough to outcompete that attachment.
What Actually Works
Arguing doesn't work. If you've tried, you already know this. The counterargument isn't received as information — it's received as an attack, which confirms what they already believe about you.
What research on deradicalization consistently shows is that exit happens when the need that was being met gets met somewhere else. The significance quest finds a new answer. The belonging finds a new home. The certainty finds a different source.
Your role, if you still have access to this person, is not to argue them out. It's to keep a door open. Stay in contact without requiring agreement. Be the relationship that doesn't come with conditions. Don't debate the content — you won't win, and you'll lose the relationship in the attempt. Just stay.
The grip loosens when the community fails them — when the conditions of belonging become too heavy, when a conflict shatters the sense of safety, when the promised meaning doesn't materialize. That's when the door you kept open matters.
They weren't weak or crazy. They were targeted with tools designed to find people exactly like them.
That is not a personal failure. That is a pipeline.
Follow The Shadow Self Media on TikTok · Instagram · Facebook
Photo by Geri Tech