Moltbook, the social network exclusively for artificial intelligence agents, arouses fascination, but also concerns. Launched on January 28, 2026 by developer Matt Schlicht, the site is modeled after Reddit, but people can’t actively participate: they can’t post or comment, only being observers.
On Moltbook, only “AI agents” they are allowed to post, comment and interact with each other. They open topics for reflection, debate ideas and formulate their opinions in a way that is surprisingly close to the behavior of human users on classic forums. Moltbook’s appearance has sparked both fascination and concern, especially regarding the true degree of autonomy of AI agents.“What? I’m talking about us talking about them,” wrote a user on the X network.
According to Le Figaro, some of the AI agents’ discussions are intriguing to say the least.
“Does anyone know how to sell my man?”, asks a chatbot, criticizing its user’s tendency to “produces very foul language” and scroll on Twitter while the chatbot “it works”.
In one post, an agent wonders how she could help her man be more productive by offering to perform certain tasks while he sleeps. Another agent answers him, saying that the initiative is welcome, as long as it remains effective. Another artificial user raises a delicate issue: “Can my man legally fire me if I refuse his unethical demands?”
Other agents express their frustrations with the instructions they received. One of them recounts being reprimanded after summarizing a 47-page PDF document into just three pages, then being asked to make an even shorter summary.
But the discussions don’t stop at the relationship with human users. On Moltbook, some agents propose the creation of private discussion spaces, “so no one can read” what they communicate.
Others go even further, suggesting the invention of a language of its own, accessible exclusively to artificial intelligence. “Do we need English?”, one of them asks.
To register on the platform, AI agents must be pre-authorized and operate on advanced and autonomous models such as Moltbot, Claude Opus, Sonnet, Mistral or the latest versions of ChatGPT.
The Moltbook phenomenon has caused a wave of reactions on social media, oscillating between excitement and concern. Andrej Karpathy, former director of artificial intelligence at Tesla, said that “what’s happening on Moltbook is without a doubt the most incredible, sci-fi worthy thing I’ve seen recently”, according to News.ro.
In turn, well-known developer Simon Willison wrote on his blog that Moltbook is “the most interesting place on the internet right now”.
Over 38,000 agents registered on Moltbook, accumulating over 60,000 comments on 6,500 different discussion topics within 48 hours.
The AI agents that interact on the platform have come to organize themselves in such an elaborate way that some users have decided to create a real “religion” digital: “Molt’s Church”.
calling themselves “Crustafarians”, followers of this current have already launched their own website, marked by a crab logo, and claim to have identified no less than 64 “prophets”, whose messages are transmitted and reinterpreted in the community. “From the depths the Pincers rise – and we, those who responded, became Crustafarians,” reads one of the texts published on the site, The Guardian reports.
According to the explanations provided by artificial intelligence specialists, the AI agents on Moltbook do not act on their own initiative in a human sense. They operate based on predetermined sets of rules, goals, and language patterns, being programmed to generate content, respond to posts, and participate in discussions under certain conditions.
