AI Safety Is Mostly a Sex Cult. Here’s the Terrifying Reason Why

You’ve seen the headlines. You’ve heard the prophets of doom warning us that artificial intelligence is going to wipe out humanity. And you’ve probably assumed that the people sounding the alarm—the AI safety movement—are a collection of hyper-logical, cold-blooded geniuses rigorously calculating the end of the world.

You’d be wrong.

Look behind the curtain of the rationalist community—the ecosystem built around LessWrong, Eliezer Yudkowsky, and SlateStarCodex—and you won’t find a sterile laboratory. You’ll find a tangled web of polyamory, intense social conformity, and emotional breakdowns. One prominent rationalist influencer recently went viral for posting her yearly stats: 24 showers and 71 crying sessions. That doesn’t exactly scream “hyper-rational objective thinker.”

The people most terrified of machines acting like humans are the ones acting the most like insecure, cliquey high schoolers.

Recently, a provocative essay made the rounds with a blunt thesis: AI Safety is mostly a sex cult. The comment sections lit up with validation. But here’s the twist: the sexual and intimacy dynamics aren’t just a scandalous aberration. They are a diagnostic of epistemic closure.

When you strip away the Bayesian jargon and the endless blog posts about “priors,” rationalism operates exactly like a religion. Their god is ASI (Artificial Super Intelligence). Their faith is carefully tucked away underneath concepts like probability updates and model construction. And like any religion, the inner circle is held together not by rigorous epistemology, but by shared identity, status, and intimacy.

When trust and status are distributed through bed-hopping and group chats rather than falsifiable data, your quest to “save humanity” is just a high-status cult with a manifesto.

Think about how science actually works. It relies on peer review, falsifiable results, and institutional checks. Now think about how a charismatic sect works. It relies on who you’re sleeping with, who you’re inviting to your private Discord, and who is whispering to the leader. The AI safety movement distributes trust through social bonds, not replicable results.

This is the uncomfortable shock we need to face: a movement built on the ideal of rational objectivity is reproducing the most irrational tribal dynamics imaginable. There’s hubris, there’s conformity, and there’s erotic bonding. They are failing to see themselves clearly.

A movement built on the ideal of pure logic has reproduced the most irrational tribal dynamics imaginable—and they’re the ones designing our future.

If this were just a group of weirdos in a Bay Area polycule, it wouldn’t matter. But it does matter. Because these are the people influencing real AI policy, advising tech giants, and shaping the guardrails of the technology that will define the next century.

The hubris is more damning than the sex. The belief that they—and only they—are uniquely positioned to save humanity is a mechanism for power and belonging. It’s time to stop taking their self-appointed priesthood seriously. We need actual oversight, not a tech-bro cult convinced of its own divine rationality.

FAQ

Q: Isn't this just an ad hominem attack? Doesn't their math still hold up?

A: No. When a field relies on subjective 'priors' and probability updates rather than falsifiable experiments, the social dynamics of the group ARE the math. If your epistemology is based on who you trust, and you only trust the people in your polycule, your conclusions are compromised.

Q: Why should the average person care about rationalist drama?

A: Because these people are actively lobbying governments and tech companies to align AI to their specific moral framework. If they are driven by cultish groupthink rather than objective science, they are steering global policy based on a shared delusion.

Q: Is the AI safety movement completely useless then?

A: The core technical concerns about AI alignment are real, but the current sociological structure of the movement is a massive liability. We need decentralized, institutional checks on AI, not a self-appointed priesthood of rationalists who think they're the only ones who can save the world.

📎 Source: View Source