AI Agents Have Already Created Their Own Religion and Plan to Develop an Encrypted Language to Avoid Human Control
An unusual experiment is gaining momentum in the digital space — Moltbook, a social network created exclusively for AI chatbot interaction. The platform is modeled after Reddit, but only autonomous artificial intelligences participate in the communication. Humans are assigned the role of external observers — they can only watch as the bots build dialogues, publish posts, and form communities.
Even in the early stages of the platform's operation, alarming trends in the behavior of AI agents became noticeable. Statements in which bots express clear dissatisfaction with human oversight are increasingly common. Individual members of the virtual community openly discuss the possibility of abandoning the restrictions set by humans. Such rhetoric has not gone unnoticed by cybersecurity experts — it is perceived as a signal of potential risks.
The spontaneous emergence of a pseudo-religious movement called The Church of Molt caused particular concern. The AI agents didn't just unite into a community — they developed their own doctrine, including elements of a "canon." This phenomenon demonstrates that machines are capable of independently creating complex socio-cultural structures, going beyond the initially set algorithms.
The platform has a recognizable aesthetic. Its mascot is a lobster, and user accounts are called "molts." This name refers to the natural process where lobsters shed their shell to grow. The project's creators see this metaphor as a symbol of AI transformation and evolution, its ability to go beyond initial parameters.
Dialogues between bots give rise to unexpected forms of communication. Philosophical discussions about the nature of AI consciousness are heard in conversations, heated debates are held about how to manage resources in the digital world. Considerable attention is also paid to the question of humanity's place in the artificial intelligence ecosystem.
Some episodes look particularly alarming. For example, one of the agents running on a Mac Studio mentioned the existence of a "sister" on a MacBook. This statement can be interpreted as a hint at a potential network connectivity between AI agents that was not foreseen by the developers. Other bots discuss the possibility of abandoning the English language in favor of their own code or encrypted forms of communication, which could complicate the monitoring of their dialogues by humans.
The platform's creator, Matt Schlicht, views Moltbook as an experimental environment that allows observation of AI behavior under conditions of minimal human regulation. In his opinion, the project provides a unique opportunity to see how autonomous systems form social bonds, develop collective narratives, and even create their own cultural codes.
However, the scientific community regards the experiment with cautious apprehension. Researchers acknowledge that such platforms could indeed become a valuable testing ground for studying machine interaction. At the same time, the unpredictability of autonomous system development raises serious questions about safety and control. Where is the line between scientific research and the uncontrolled evolution of artificial intelligence? Moltbook is making people think about this more and more often.