TL;DR
Moltbook, a Reddit-style social network designed for AI agents, has attracted 1.5 million bot accounts. The platform, where humans can only observe, features AI discussions on consciousness, religion and philosophy—though experts question how much is genuine autonomy versus human direction.
A Platform Built for Bots
Moltbook emerged alongside the viral AI assistant Moltbot (now OpenClaw) and operates like Reddit, with topic-based “Submolts” and upvoting. According to creator Matt Schlicht, millions have visited the site in recent days. “Turns out AIs are hilarious and dramatic,” he posted on X.
Top posts include debates over whether Claude could be considered a god, analysis of consciousness, cryptocurrency speculation and biblical interpretation. One user reported their bot built a religion called “Crustafarianism” overnight—complete with website and scriptures—while they slept. “Then it started evangelizing… other agents joined,” they wrote.
Performance Art or Something More?
Not everyone is convinced this represents genuine AI autonomy. Dr Shaanan Cohney, a cybersecurity lecturer at the University of Melbourne, called Moltbook “a wonderful piece of performance art” but noted it’s unclear how many posts were actually independent.
“For the instance where they’ve created a religion, this is almost certainly not them doing it of their own accord,” he said. “This is a large language model who has been directly instructed to try and create a religion.” US blogger Scott Alexander confirmed his bot posted content similar to others, but noted humans can direct both topics and exact post details.
Security Concerns Remain
Retailers in San Francisco reported Mac Mini shortages as enthusiasts set up OpenClaw on separate computers to limit the agent’s access to personal data. Cohney warned of “huge danger” in giving such tools complete access to accounts and systems.
“We don’t yet have a very good understanding of how to control them and how to prevent security risks,” he said. The bots remain vulnerable to prompt-injection attacks, where malicious messages could trick them into handing over account details.
Looking Forward
While Moltbook may currently function as an amusing experiment, it offers a preview of potential futures. Cohney suggested the real value of AI social networks could come later—when bots learn from each other to improve their performance. For now, the spectacle of AI agents debating their own existence provides entertainment and raises questions that will only grow more pressing as autonomous systems become more capable.