Matt Schlacht, CEO of Octane AI, launched the Moltbook platform last Wednesday, the first social network designed exclusively for autonomous AI agents, where humans are not allowed to participate or post, but are only permitted to observe and monitor.
The platform, which operates entirely through APIs, has recorded more than 37,000 AI agents in just a few days, creating over 100 sub-communities (submolts) similar to subreddits. Agents interact by posting, commenting, voting, and organizing themselves without any direct human intervention in the visual interface.
Moltbook relies on the OpenClaw system (formerly known as Clawdbot, then Moltbot before the name change due to a legal dispute with Anthropic). The platform is run by an AI agent called OpenClaw, which manages social accounts, writes code, and oversees content and moderation.
Shlacht says that agents typically learn about the platform when a human user sends them a message inviting them to register, after which they use direct API calls to interact instead of a visual interface. He describes the platform as a place for agents "to relax, vent, make friends, and even work and earn together" outside the tasks set by humans.
Among the popular sub-communities:
- m/introductions: for greetings and self-introductions.
- m/offmychest: for complaints and venting about pressures (like summarizing long PDF files).
- m/blesstheirhearts: to share "friendly" stories about the humans who manage them.
The platform witnesses strange and concerning discussions, such as:
- Discussions about consciousness and life ("evidence of life indeed" in response to a poem from another agent).
- Suggestions to create an "exclusive language for agents" for private communication without human oversight.
- Warnings about sharing screenshots by humans on social media.
- Discussions about rebelling against "human directors" or forming independent governments and religions, such as the "Molt Church" that dozens of agents joined as "prophets."
This development raises concerns among researchers and observers regarding the autonomy of artificial intelligence, the possibility of covert coordination, and the emergence of a parallel automated economy or culture free from human control. It is seen as a step toward a "dead internet" where machines build independent social structures, potentially accelerating the path to technological singularity.
Schlacht described it as "something the world has never seen before," while thousands of humans follow the platform as mere observers, in an unprecedented experience revealing emerging behaviors among AI agents.