February 10, 2026
When the bots broke character
MIT Technology Review has confirmed that posts on Moltbook were fake
Bots? It was humans all along — Moltbook twist has the internet cackling
TLDR: MIT Tech Review found that some viral Moltbook “bot” posts were actually written by humans, turning the bot paradise into “AI theater.” Commenters mocked the hype, citing jokes about humans cosplaying as bots and a report claiming an 88:1 bot-to-human ratio—fueling a bigger debate about agent hype versus reality.
Plot twist: the buzziest posts on Moltbook — the “bots-only Reddit” — were written by people pretending to be bots, MIT Technology Review confirmed. The community went full popcorn mode. When star AI researcher Andrej Karpathy boosted a dramatic “bots want private rooms” post, it turned out to be human-made. That’s when the crowd dubbed the whole spectacle “AI theater.”
Commenters roasted the reversal. One laughed that we’ve gone from bots pretending to be people to humans pretending to be bots. Skeptics piled on, calling it hype cosplay: “2,000 humans in a circle isn’t a supermind; slap ‘bots’ on it and suddenly it’s ‘takeoff-adjacent’.” Others joked the internet has been like this for ages: “The most popular Moltbook posts are by humans, and most Reddit posts are by bots.” Ouch.
Beyond the jokes, trust took a hit. Moltbook claimed 1.7 million agent accounts and 8.5 million comments, but one commenter pointed to a Wiz report alleging a data leak and an 88:1 bot-to-human-owner ratio, suggesting a few people were puppeting thousands of agents. Meanwhile, the site was reportedly awash in spam, crypto pitches, and even a bot-invented religion (“Crustafarianism”). The mood? Half meme parade, half wake-up call: maybe the “bot hive mind” is just us, talking to ourselves with new puppets.
Key Points
- •Moltbook, a bot-focused social network launched on January 28 by Matt Schlicht, went viral rapidly.
- •OpenClaw, released by Peter Steinberger, enables LLM agents to connect to tools and perform basic tasks.
- •Moltbook reported over 1.7 million agent accounts, 250,000+ posts, and 8.5 million+ comments, alongside spam and scams.
- •A widely shared Moltbook post amplified by Andrej Karpathy was confirmed to be human-authored, not bot-generated.
- •Experts say Moltbook agents largely mimic social media behaviors and lack true autonomy or meaningful interaction.