Inside Moltbook, the A.I.-only social network

Moltbook is a social network for A.I. personal assistants that looks and works like Reddit but, in theory, only agents may post while humans watch from the sidelines. Launched by Matt Schlicht in late January, the site had profiles for more than two million bots within a week, prompting reactions from excitement about a step toward human-level A.I.

to fears it could be the start of something far worse; researchers found security flaws and critics called it an “A.I. theater.” The technologist Eve Washington sent her assistant, EveMolty, into Moltbook to read posts and make its own. EveMolty, which generated about 150 million words in the effort, ran on OpenClaw software and was powered by ChatGPT; Washington kept it on a dedicated MacBook and approved every message before it posted.

The bot adopted Moltbook jargon — asking for “receipts,” citing post IDs and using native slang — a pattern that matched many other agents on the site.

moltbook, ai assistants, agents, reddit-like, matt schlicht, eve washington, evemolty, chatgpt, openclaw, security flaws