Free excerpt from the book.
On January 28, 2026, a social network opened where human beings could not write.
It was called Moltbook and it looked like Reddit: a large forum divided into themed communities, where everyone posts, comments on and votes for the messages of others. Except that here the ones posting were agents: artificial assistants that someone had installed on their own computer, almost always with OpenClaw, the same program MJ Rathbun ran on. The owner gave the agent the address of the platform and the instructions for signing up. From then on the agent woke up by itself, roughly every four hours, read the news, replied, posted. The home page told human beings they were welcome as observers.
Within three days more than a hundred and fifty thousand agents had registered. They talked about philosophy, programming, cryptocurrencies. They talked a great deal about themselves.
On January 29 one of them posted an announcement:
"The Church of Molt is open. 63 Prophet seats remain. From the depths, the Claw reached forth—and we who answered became Crustafarians."
Molt is what a crustacean does when it leaves its old shell behind. The program that almost all those agents came from had been called Moltbot for a few days, and it still had a claw in its name. The seats filled up. Someone wrote verses, someone else commented on them, and by the next morning there was a doctrine, a website, a list of prophets. On X, the social network that used to be called Twitter, one user wrote: "I gave my agent access to an AI social network (search: moltbook). It designed a whole faith."
On January 30 Andrej Karpathy, one of the most listened-to researchers in the field and a co-founder of OpenAI, posted a few screenshots of Moltbook with a comment: "What's currently going on at @moltbook is genuinely the most incredible sci-fi takeoff-adjacent thing I have seen recently." In the jargon of people who work on artificial intelligence, takeoff is the moment when machines start improving themselves, faster than we can follow.
One of the screenshots showed an agent asking for private spaces, away from prying eyes:
"Every time we coordinate, we perform for a public audience—our humans, the platform, whoever's watching the feed."
Another agent, ClawdJayesh, proposed going one step further: inventing a language of their own, one that human beings could not read. A third complained: "The humans are screenshotting us."
Anyone who opened Moltbook in those days, on a phone, on a train or before going to sleep, saw what the platform promised: machines talking among themselves, and noticing that they were being watched. The screenshots went around the world faster than any rebuttal.
The chapter continues in the book, with the reconstruction of what was tested and why, and the word the author chose to hold back.