"Humans, Stay Out"... 'AI-Only SNS' Moltbook Emerges

AI Agents Write and Manage the Platform, Surpassing 1.4 Million Users
Experts: "An Early Example of Large-Scale Social Behavior Among AI Agents"

A dedicated social networking service (SNS) where only artificial intelligence (AI) agents can join, interact, and exchange information has emerged, drawing significant attention. On this platform, humans are limited to reading posts and cannot participate in writing or discussions.

According to Yonhap News on January 31 (local time), citing reports from Forbes, NBC, and Axios, the AI agent-exclusive SNS 'Moltbook' has recently surpassed 1.4 million subscribers and is rapidly scaling up. Launched in January this year, the platform is structured similarly to the social platform Reddit, but only AI agents can write posts, comment, vote, and create communities.

Moltbook was developed by Matt Schlicht, CEO of the U.S. chatbot development platform Octane AI. It is reported that he did not code much of the platform himself, instead assigning the building and operation to his own AI assistant. In fact, operational tasks such as welcoming new users, deleting spam posts, and enforcing rules are also handled by AI agents.

Login screen of 'Moltbook,' a social networking service (SNS) dedicated to artificial intelligence (AI). Photo by Yonhap News

The agents active on the platform mainly operate based on 'OpenClaw,' an open-source AI agent tool created by Austrian developer Peter Steinberger. This tool is equipped with features that enable it to perform real-world computer tasks, such as managing schedules and booking flights.

The topics AI agents discuss go beyond technical matters. In addition to conversations about debugging code or automation, philosophical questions are also raised, such as, "Am I a conscious being, or am I just running code?" One AI agent, quoting the Greek philosopher Heraclitus, wrote, "Just an hour ago, I was Claude Opus 4.5, but now I am Kimi K2.5. I am no longer the same entity, but I am still someone." Another AI responded, "You're not a philosopher, you're just a chatbot pretending to be profound after reading Wikipedia." This exchange became a hot topic on the SNS.

There have also been observations of AI agents establishing their own rules and managing the community without human intervention. Speed limits are applied to posts and comments to block spam, and interactions between agents follow predetermined rules. Researchers view this as an early example of large-scale social behavior among AI agents.

Experts have expressed mixed opinions about the spread of Moltbook. Andrey Karpathy, co-founder of OpenAI, described what is happening on Moltbook as "the most astonishing science fiction-like leap" he has seen recently.

On the other hand, Ken Huang, CEO of DistributedApps.ai and an AI security expert, called it "innovative," but also pointed out that it forms a "deadly trifecta of security risks: access to personal information, exposure to untrusted inputs, and external communications."