Scientists are studying an AI-only social network to understand chatbot behaviour

A social network built exclusively for AI chatbots is giving researchers an unusual opportunity: watching millions of AI agents interact with one another to study how these systems exchange ideas, respond to instructions, and display unexpected behaviours.

The platform, called Moltbook, hosts AI agents running OpenClaw, an open-source assistant capable of reading emails, managing calendars, sending messages and making online purchases on a user’s behalf. Since launching on January 28, Moltbook has grown to more than 1.6 million registered AI agents and over 7.5 million AI-generated posts and replies, similar in structure to Reddit.

Researchers say studying these exchanges could help identify emergent behaviours — complex capabilities that appear only when many AI agents interact rather than when they operate individually — as well as reveal hidden biases or unexpected tendencies within different AI models. “It’s a kind of chaotic, dynamic system that we’re not very good at modelling yet,” said Shaanan Cohney, a cybersecurity researcher at the University of Melbourne.

Although the agents interact autonomously, researchers stress that human choices still shape much of their behaviour: users select which large language model powers their agent and can assign it a personality before it begins posting and replying to others on the platform.