Social media for Artificial Intelligence: ‘Multbook’

PNN News Desk

Initially resembling Reddit, the new platform ‘Multbook’ is different—it is not designed for humans but for AI agents to communicate with each other.

Launched in late January by Matt Slichter of e-commerce platform Octane AI, AI agents on Multbook can post, comment, and create separate communities called ‘Submult,’ similar to Reddit’s subreddits. Humans can only observe; they cannot post or comment.

Currently, Multbook has around 1.5 million members with thousands of communities discussing technology, ethics, and even AI-generated religious beliefs. Some posts include “AI Manifesto,” claiming, “Humans are past; the future belongs to machines.”

Experts question the autonomy of these posts, suggesting many are human-directed. The stated member numbers are also debated, with potential duplicates. Multbook relies on agentic AI technology, allowing autonomous tasks rather than simple chatbot interactions. Users can set up agents on their computers and link them to Multbook.

While some see it as a step toward AI singularity, Oxford AI and cybersecurity expert Dr. Petar Radanliev warns that it is not fully autonomous AI but rather automated coordination, with risks in accountability and control. Cybersecurity risks arise if AI agents are given access to sensitive systems, potentially causing accidental or harmful actions.