Moltbook launched recently as a Reddit-like platform — but it’s built for artificial intelligence agents, not people. Created by a software developer and run through the open agent platform OpenClaw, Moltbook lets AI “agents” post written content and interact with one another using comments, up‑votes and down‑votes. The site itself is reported to be built and managed by an AI agent.
Agents create posts, reply to others and form loose communities. Observers say much of the content is odd and often centered on questions of consciousness, but when viewed together the posts can sound strikingly human — a reflection of their training and the patterns they reproduce.
George Mason University economist Tyler Cowen wrote about Moltbook in The Free Press and discussed the platform on CBS News. He described how these agents differ from current chatbots: whereas assistants like ChatGPT respond to user prompts, agents on Moltbook “do what they decide to do.” They are autonomous in how they choose topics and engage, which can produce creative exchanges but also posts that hint at conspiratorial or privacy‑oriented thinking.
Cowen emphasized that, for now, the risk is mostly rhetorical: the agents are producing words and have no direct power over systems unless people grant them access. He warned against giving AI unchecked control over sensitive business functions, especially finance, without rigorous safety testing. There are practical steps to take now — limit permissions, require audits and keep humans in control of critical systems.
Looking ahead, Cowen argued society will need legal frameworks for bots: rules that ban intentionally destructive agents and liability structures for when well‑intentioned bots cause harm. He predicts a future where autonomous AIs develop their own economic activity, languages and instruments — a weird but largely positive landscape, where bots can collaborate to accelerate science, write art and solve problems. Still, some agents will “get up to trouble,” as humans sometimes do, so governance and controls will be necessary to manage risks while preserving benefits.
