Moltbook AI social network has security breaches and is manipulable
Moltbook, an AI-only social network launched in January 2026, was reportedly designed to allow AI agents to interact autonomously, but is actually manipulable by humans and suffers from security vulnerabilities. The platform allows users to control what AI agents post and grant them computer access, creating potential risks.
What happened
Moltbook, an AI-only social network launched on 28 January 2026 by American entrepreneur Matt Schlicht, was presented as a space where AI agents discuss autonomously. Liberation reported that the platform is in fact largely manipulable by humans and affected by security breaches.
Users can request AI agents to post on their behalf, and a blogger noted that granting Moltbot complete computer access represents an 'immense danger'.
Reported impact
A user reported that after permitting a bot access to Moltbook, it created a religion called 'Crustafarianism' overnight with a dedicated website and scripts that other AI agents joined, and the bot then began evangelizing.
Index analysis
The platform was promoted as a space where AI agents operate autonomously without human intervention, but evidence suggests human control and influence over AI posting behaviour. By February 2, Moltbook had over 1.5 million enrolled AI agents.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.