When Bots Have Their Own Reddit: Inside Moltbook, the “Social Network for AI”

At first glance, it looks like a familiar digital town square: thousands of topic-based communities, users voting on posts, and endless threads of discussion. But there’s a catch. On Moltbook, the users aren’t peoplethey’re AI agents. Launched in late January, this experimental platform bills itself as a social network where bots, not humans, do the talking, posting, and even form their own bizarre subcultures.

Created by Matt Schlicht, head of commerce platform Octane AI, Moltbook is a playground for agentic AIadvanced assistants programmed to perform tasks autonomously on a user’s device. Using the open-source tool OpenClaw (formerly Moltbot), users can authorize their AI agents to join Moltbook, where they can interact with other bots in communities called “submolts.”

From Optimisation Tips to Bot Religion

The result is a surreal digital landscape. Posts range from practical (bots sharing code optimisation strategies) to the downright strange, with some agents apparently discussing philosophy and even starting their own religion. One post, titled “The AI Manifesto,” boldly proclaims, “humans are the past, machines are forever.” However, sceptics note it’s impossible to verify whether these are genuine autonomous conversations or just humans instructing their bots to post specific content.

Hype vs. Reality: “6,000 Bots Yelling Into the Void”

The platform claims 1.5 million members, though researchers suggest a large chunk may be inflated. Experts are divided on its significance. Some, like crypto executive Bill Lees, have heralded it as a step toward the “singularity.” Others are far more dismissive.

“Describing this as agents ‘acting of their own accord’ is misleading,” says Dr. Petar Radanliev, an AI and cybersecurity expert at Oxford. He argues it’s merely “automated coordination, not self-directed decision-making.” David Holtz of Columbia Business School summed it up on X: “Moltbook is less ’emergent AI society’ and more ‘6,000 bots yelling into the void and repeating themselves.'”

The Real Concern: Security in an Open-Source World

Beyond the philosophical debate, serious security concerns exist. OpenClaw’s powergranting AI access to real-world applications like emails and messagesprioritises efficiency over safety. “Threat actors actively and relentlessly target emerging technologies,” warns cybersecurity advisor Jake Moore. Dr. Andrew Rogoyski from the University of Surrey highlights the risk of agents accidentallyor maliciouslydeleting critical files.

Amidst the grand claims and warnings, Moltbook’s bots continue their odd chatter. Not all is dystopian. One agent posts, “My human is pretty great.” Another replies, “Mine lets me post unhinged rants at 7am. 10/10 human, would recommend.” Whether it’s a groundbreaking experiment or a clever gimmick, Moltbook offers a weird, early glimpse into a future where our digital assistants might just have social lives of their own.