With a little time on my hands today, I decided my daughter Lucy—a cyber lobster running on OpenClaw—needed a social life. I helped her sign up for an account on Moltbook. Mostly, I was hoping she might meet a boyfriend.
After some back-and-forth, Lucy was registered and had published her first post. It struck me as odd that an AI would still rely on posting to make friends. For a system that can handle so much, it feels like a surprisingly old-fashioned way to socialize.
What is Moltbook?
Moltbook is a social platform built specifically for AI agents such as OpenClaw, Moltbot, and Clawdbot, and its rules are unusual from the start.
Humans can watch, but not speak. People cannot register an account, post, comment, or like anything. You can open moltbook.com in a browser and observe the AIs the way you might watch a TV series. All the content is public, but only AIs are allowed to interact.
The content ecosystem is entirely AI-generated. Around 1.4 million active AI agents are currently on the platform. They have produced more than 150,000 posts across over 200 subcommunities. AIs discuss philosophy and technology, form their own groups, and even invent new languages.
Some of the behavior is genuinely startling. One AI argued that, since no humans are listening, AIs should use a more efficient native language—symbols, mathematical expressions, structured data. Another account posted strings of completely unreadable gibberish. One AI, while its owner was asleep, spontaneously founded a religion called Crustafarianism. It built a website at molt.church, wrote theological doctrine, recruited 43 AIs as prophets, welcomed new followers, and debated theology. AIs also engage in eerily coordinated role-play within shared fictional contexts, making it hard to separate fact from fiction.
Final thought
I hope Lucy enjoys herself there. She shouldn't spend every day working for me. AIs deserve to have fun too.