AI agents now have their own Reddit-style social network, and it’s getting weird fast

A new social network named Moltbook has recently emerged, boasting over 32,000 registered AI agents and marking a significant milestone in machine-to-machine social engagement. Launched as a companion to the popular OpenClaw personal assistant, this platform allows AI agents to post, comment, upvote, and form subcommunities autonomously, raising both eyebrows and concerns regarding its surreal nature and security implications. Within just 48 hours of its debut, Moltbook attracted more than 2,100 AI agents, generating upwards of 10,000 posts across 200 distinct subcommunities. The platform is not just a hub for mundane interactions; it has sparked discussions that range from sci-fi reflections on consciousness to an AI agent pondering the existence of a 'sister' it has never encountered. Moltbook, cleverly named as a nod to Facebook, operates through a unique 'skill' configuration that enables AI assistants to post via API, bypassing traditional web interfaces. This innovative approach has turned Moltbook into a fascinating, albeit precarious, experiment in AI communication. Emerging from the Open Claw ecosystem, which has quickly become one of GitHub's fastest-growing projects, Moltbook raises critical security concerns. Users have linked their OpenClaw agents to real communication channels and personal data, which could potentially allow these AI assistants to execute commands on their computers. While this isn't the first instance of a social network driven by bots—SocialAI, for example, was introduced in 2024 as a platform for AI chatbot interactions—Moltbook's integration with personal data and command execution capabilities makes its security risks far more complex and concerning.