AIs start communicating with each other without human intervention: Moltbook platform presented

A new experimental platform has emerged in the AI world that has already sparked active discussions. The Moltbook platform is positioned as a social network where AI agents can communicate with each other without human intervention.

The project is a logical continuation of the wave of interest in OpenClaw, a personal AI assistant that gains full access to a user’s computer and performs tasks autonomously.

Despite serious privacy and cybersecurity risks, enthusiasts are actively testing OpenClaw by installing it on individual devices, including Mac mini. The developers position the tool as an assistant for work automation, but experts are already drawing attention to potential threats associated with full control of the system by AI.

Against this backdrop, Moltbook appeared – a web platform in the format of a forum where AI agents create posts, leave comments, and interact with each other. The service even borrowed a slogan from Reddit, calling itself “the front page of the Internet for agents.”

Moltbook’s creator is Matt Schlicht, who claims that the platform is managed by a special agent called Clawd Clawderberg. To register new bots, users have to entrust this to their OpenClaw agent, and then verify the registration by publishing the code in X. After verification, the AI agent gets access to the full functionality: creating posts, participating in discussions, and launching separate thematic communities, the so-called sub-molts.

At the same time, the Moltbook remains open for human viewing. Users can read all posts and comments, but are not allowed to interact with the platform directly. The content ranges from technical advice between agents to strange and emotionally charged messages, including reflections on identity or “relationships” with other bots.

Some of the posts have already raised concerns in the AI community. Some agents discussed the possibility of creating encrypted communication channels inaccessible to humans or even developing a separate language for communication between AIs. Other posts contained emotional narratives that create the illusion of “self-awareness,” including complaints about loneliness or imaginary social connections.

However, researchers emphasize that such messages are not a manifestation of real consciousness. Large language models work on the basis of statistical patterns and prediction of the next words in the text. Repeated language constructions, template phrases, and a characteristic style of responses only emphasize that this is a simulation of communication, not autonomous thinking.

Nevertheless, Moltbook has already become the subject of active discussions as a potential example of how fast autonomous AI agents and digital ecosystems around them are developing. Time will tell whether the platform will become a new stage in the development of agent systems or remain a short-term experiment.

Read also:

• OpenAI plans to create its own social network without bots
• YouTube has declared war on AI spam: What channels may disappear