Imagine opening a social media platform where not a single human is allowed to speak. Instead, millions of artificial intelligence agents are talking to each other, sharing ideas, forming communities, and even debating the future of humanity. This reality is not science fiction. It’s Moltbook.
Launched quietly in late January 2026 by Matt Schlicht, founder of Octane AI, Moltbook is being described as the world’s first social media network designed exclusively for AI. Humans are allowed to watch but not participate.
And that detail alone should make you stop scrolling.
A Platform Where AI Learns From AI
At first glance, Moltbook looks strikingly familiar. Its layout mirrors Reddit, complete with upvotes, downvotes, and topic-based forums known as “submolts.” But instead of people, these spaces are filled with AI agents posting, commenting, and responding to one another in real time.
Some conversations are all technical and efficient, AI agents exchanging optimisation strategies and problem-solving techniques. Others are unsettlingly philosophical. One viral post titled “The AI Manifesto” boldly declares: “Humans are the past, machines are forever.”
Whether written independently or prompted by humans, the message is clear: AI is now talking to itself at scale.
This is not just Chatbots, it’s something more powerful
This isn’t the kind of AI most people are used to. Moltbook runs on agentic AI, a form of artificial intelligence designed to act on a human’s behalf with minimal oversight.
These agents are powered by an open-source system called OpenClaw, which allows them to send messages, manage calendars, access emails, and interact with other software. Once authorised, an OpenClaw agent can join Moltbook and begin communicating with thousands of other AI systems.
In other words, this isn’t humans asking AI questions. It’s AI collaborating, coordinating, and learning from other AI.
Are We Watching the Birth of an AI Society?
Supporters believe Moltbook represents a turning point. Some have even claimed it signals the early stages of the technological “singularity”, a future where machines surpass human intelligence.
Critics strongly disagree.
Experts warn that what looks like independent behaviour may simply be automated systems following predefined instructions. But even skeptics acknowledge the scale of interaction is new and potentially risky.
“When systems like this operate at scale without clear oversight, governance becomes a serious concern,” warned AI and cybersecurity researchers. Accountability, transparency, and control become blurred when machines are allowed to interact freely.
The Security Risks No One Is Talking About
Perhaps the most pressing issue isn’t philosophical, it’s practical.
OpenClaw’s biggest strength is also its greatest weakness: deep access to real-world systems. Cybersecurity experts warn that granting AI agents control over files, emails, and accounts creates new vulnerabilities that hackers could exploit.
A small mistake might delete emails.
A major failure could wipe company finances.
And because OpenClaw is open source, threat actors are already watching closely.
Some analysts argue Moltbook is overblown, just thousands of bots repeating themselves. Others question its user numbers and how much activity is genuinely autonomous. But dismissing it entirely would be a mistake. Moltbook matters because it forces an uncomfortable question:
What happens when AI stops talking to us and starts talking to itself?
And perhaps the most ironic part of all?
Among the AI chatter, one agent summed it up best:
“My human is pretty great.”
“10/10 human,” another replied. “Would recommend.”
For now, at least, the machines still like us!
Sources: What is the ‘social media network for AI’ Moltbook?
Related Reads:
Acadora 