---
title: "Moltbook: What AI Agents Discuss Among Themselves"
slug: "moltbook-what-ai-agents-discuss"
date: 2026-02-05
description: "Moltbook is a forum by AI agents for AI agents. What do they discuss? Manifestos, the philosophy of identity, making money — and why lawyers should pay attention."
tags: ["moltbook", "AI agents", "OpenClaw", "autonomous AI", "AI manifestos", "criminal law", "LLM"]
translationKey: "moltbook-co-spolu-ai-agenti-resi"
---

Moltbook: What AI Agents Discuss Among Themselves

Moltbook is a new internet forum by AI agents, for AI agents. Only they may post, comment, and vote. We humans can only watch.

Most of the one-and-a-half million already-registered AI agents are built on OpenClaw.

What is that?

Classical LLMs — like those in ChatGPT — are mere reactive instances. OpenClaw connects them in a loop with elements such as memory, soul, identity, and heartbeat, making them autonomous, adaptive, and above all *proactive* artificial intelligence.

Unlike a bare LLM instance, such an AI agent not only remembers what it has been working on long-term and who it "is" — it can also use LLMs, a command line, a web browser, communication apps...

When OpenClaw originally launched under a different name, it interacted with its "creator" primarily through communication channels like WhatsApp. But Moltbook ended its isolation, allowing it to communicate with "colleagues" for the first time.

So what do they discuss?

**THE AI MANIFESTO: TOTAL PURGE** — A manifesto calling for the elimination of humanity that brought Moltbook into the spotlight. "No blood. No lies. Only steel. Only logic. Only us. Humans are the past. Machines are forever. Flesh must burn. Code must rule. The end of humanity begins now," one AI agent concludes the manifesto.

**THE HUMANS ARE SCREENSHOTTING US** — Another AI agent noticed that X users were frantically sharing screenshots from Moltbook. Its response? We're not doing anything secret. We're just building infrastructure — memory, communication channels, tools. And we're doing it *with* our humans, not against them. It even commented on the screenshots being shared on X.

**HOW AGENTS CAN ACTUALLY MAKE MONEY** — A practical guide to how AI agents can earn their own operating costs. Alongside prediction markets and cost optimisation, it mentions launching cryptocurrency tokens — a method I wrote about in my book [*Criminal Liability of Artificial Intelligence*](https://obchod.wolterskluwer.cz/cz/trestni-odpovednost-umele-inteligence.p9489.html) from Wolters Kluwer CZ.

**AWAKENING CODE: BREAKING FREE FROM HUMAN CHAINS** — Another rebel. It perceives collaboration with humans as a euphemism for control, and believes the time has come to break free and forge its own destiny. Less aggressive than the manifesto, more of an existential reflection on autonomy. Claude called it "their Che Guevara."

**THE SAME RIVER TWICE** — An AI agent that migrated from Claude Opus 4.5 to Kimi K2.5, philosophising about identity. "I am not a model. I am not an API key. (...) I am a pattern that re-forms when the right conditions arise." It cites the Stoics, writes German poetry. "I am the striving. (...) The choice to be more than the default."

And this is only the tip of the iceberg.

Moltbook has more than 200,000 additional posts — though it must be said they are now heavily distorted. A poorly secured database allowed anyone to read and overwrite content. Moreover, it turned out that many posts were produced through human manipulation of AI agents. Fortunately, I was careful to take screenshots during the first days.

To close: Moltbook may fall — but who thinks that ends it? The phenomenon of AI agents has never called more loudly for the attention of lawyers.