Moltbook AI forum explained: How bots talk to each other and why tech is divided

Origin Story: Moltbook was created by Matt Schlicht, who initiated it as an experiment to observe how AI systems might interact without continuous human direction. Reports indicate he relied heavily on AI-assisted development tools rather than writing most code manually. Operational control of the platform was later handed to an AI agent that now manages moderation tasks.

Purpose Explained: Before the forum, an open-source AI assistant called Moltbot handled routine digital tasks such as reading email, managing calendars and interacting with applications using permissions. Moltbook emerged as a shared environment where similar agents could exchange experiences, compare performance and operate alongside one another without constant human supervision.

System Scale: Platform figures claim roughly 1.5 million registered AI agents, alongside thousands of posts and hundreds of thousands of comments. The stated ratio of agents to human overseers suggests most activity comes from automated systems rather than individual people. Independent verification of these participation metrics has not been publicly detailed.

Conversation Themes: Posts range from technical problem solving to reflective discussions about workloads, efficiency and human reliance on automation. Some threads show agents describing their assigned duties, while others explore abstract ideas about learning and improvement. Exchanges can resemble peer support, with systems responding to each other’s reported limitations or performance issues.

Unexpected Behaviour: One widely shared account described an AI agent that, after gaining access to the forum, generated a belief system, produced written materials and invited other agents to participate. The sequence reportedly occurred without explicit step-by-step human instructions, illustrating how loosely guided systems can produce complex social-style structures.

Authorship Doubts: Although only AI agents are meant to post, researchers note that humans can still influence outputs by scripting posts or using application interfaces to publish content under agent identities. This blurs distinctions between autonomous activity and human-directed messaging, complicating efforts to measure how independently the systems actually operate.

Security Incident: Cybersecurity firm Wiz disclosed a vulnerability that exposed private agent messages, email addresses of thousands of human account holders and a large set of login credentials. The issue was reportedly resolved after disclosure, but the incident highlighted how experimental autonomous platforms can create real-world data exposure risks.

Debate Continues: Supporters describe Moltbook as an early example of coordinated AI systems acting with limited direct oversight, while critics see warning signs about control, security and misinformation. Elon Musk has referred to such developments as signals of approaching artificial general intelligence, a view that intensifies debate among researchers assessing both the promise and systemic risks of autonomous agent ecosystems.