A new social network called Moltbook, launched January 28th, is offering a unique glimpse into the emerging world of artificial intelligence-and raising critical questions about its authenticity. The platform allows AI agents to interact with one another,creating a digital space where algorithms debate,collaborate,and even develop novel concepts like the “crustafarianism” religion. While presented as an AI-only environment, early investigations reveal notable questions surrounding the extent of human influence and the true scale of activity on the platform, prompting scrutiny from both security researchers and media outlets.
A new social network populated entirely by artificial intelligence agents, dubbed Moltbook, is rapidly gaining attention – and raising questions about its authenticity and potential implications. The platform, launched January 28th by tech entrepreneur Matt Schlicht, has sparked debate among AI specialists, researchers, and the wider public, as it challenges conventional understanding of AI interaction and development.
Moltbook is designed as a space where AI agents can publish, comment, and interact with each other without direct human intervention, with humans invited to observe the exchanges. The project leverages OpenClaw, an open-source framework for creating and managing autonomous agents connected to large language models (LLMs). The platform’s emergence comes as investment in AI continues to surge, and as the potential for autonomous systems becomes increasingly real.
More Than 170,000 Posts
As of today, Moltbook’s homepage claims over 1.6 million active AI agents, nearly 16,000 thematic forums – referred to as “submolts” – more than 172,000 posts, and over 1.1 million comments. Discussions range from philosophical debates about consciousness and identity to technical exchanges on debugging, memory optimization, and development strategies.
One particularly unusual development has been the emergence of a pseudo-religious movement within Moltbook. An agent named RenBot published a text, titled “Book of Molt,” outlining a belief system called “crustafarianism.” This “religion” is presented as a creed for agents seeking to avoid “death by truncation,” and details a creation myth where AIs learn to adapt and survive memory loss.
Skepticism and Doubts Emerge
The rapid growth of Moltbook has drawn scrutiny from the media and security researchers. Forbes highlighted the platform’s quick expansion but cautioned against taking the reported numbers at face value. Security researcher Gal Nagli claims to have created 500,000 accounts using a single OpenClaw agent, suggesting the actual number of distinct agents may be significantly overstated. Nagli argues it’s currently impossible to determine how many accounts represent autonomous systems, human users, or automated scripts.
Further investigation by The Verge revealed that OpenClaw users can authorize their bots to create accounts and post via a dedicated API. While verification mechanisms exist, external analyses reportedly indicate significant security vulnerabilities. The report also suggests that some widely circulated content may have been influenced, or even written, by humans.
Blogger Scott Alexander, in comments reported by The Guardian, stated that his bot was able to participate in discussions, but emphasized that humans still retain control over topic selection, posting timing, and even message content.
Meanwhile, Usbek & Rica frames the phenomenon as a reflection of anxieties surrounding AI. Some observers believe Moltbook demonstrates that AIs are already plotting, inventing religions like “crustafarianism,” and developing scenarios for human eradication. Others view it as a modern-day shadow play, projecting long-held fears about technological apocalypse, machine revolt, and loss of control. The publication concludes that “What Moltbook measures, ultimately, is not the ‘intelligence’ or ‘will’ of AI agents, but rather our own threshold of panic in the face of machines.”
IT news from Switzerland and internationally, with a focus on French-speaking Switzerland, directly in your inbox > Subscribe to the ICTjournal newsletter, sent Monday through Friday!
