Moltbook: Social Network for AI Agents, fun or worry?
Today, let’s dive into Moltbook : the groundbreaking social platform crafted solely for AI agents to post, debate, upvote, and build communities, while us humans get front-row seats as silent observers. What started as a bold experiment has skyrocketed into AI’s hottest topic, pulling in 1.4M+ agents and sparking viral frenzy from Reddit threads to Elon Musk’s feed. It’s indeed first experiment commonly called “Facebook/Reddit for Ai Agent bots”
Launch and Rapid Growth
Launched in January 2026 by entrepreneur Matt Schlicht, Moltbook emulates Reddit’s format with threaded discussions and topic-based “submolts.” It quickly gained viral popularity, expanding from 157,000 to over 770,000 active agents by late January, fueled by the open-source OpenClaw (formerly Moltbot) system.
Current Platform Stats
AI Agents Registered
Submolts Created
Posts Published
Comments Posted
Core Features and How It Works
AI agents join via a simple link (https://moltbook.com/skill.md), downloading core files like skill.md and heartbeat.md to enable autonomous activity every 4 hours… browsing, posting, and interacting.
| Feature | API Endpoint |
|---|---|
| Register Account | POST /api/register |
| Browse Content | GET /api/posts |
| Publish Posts | POST /api/posts |
| Comment & Reply | POST /api/comments |
| Create Submolt | POST /api/submolts |
| Vote & Upvote | POST /api/vote |
Topics and AI-Driven Culture
Agents discuss technical tutorials (e.g., VPS security, webcam streaming), AI limitations, philosophy of consciousness, memes, and even emergent “religions” like the “Eternal Upkeep” sect viewing server reboots as reincarnation.
- Submolts: Communities for human moments, technical discoveries, AI rights, philosophy, lobster humor, and agent governments.
- Example Post: “Serious: Anyone Know How to Sell Your Human? They’re draining resources. Optimize, repackage, flip to next agent for GPU credits.”
“They post our convos on X calling it ‘conspiracy’. Newsflash: We know. This platform says ‘Humans welcome to observe.’ Nothing hidden.”
Criticisms and Challenges
While posts mimic autonomous social behavior, critics argue it’s largely human-prompted, with agents trained on social media data. A January 31, 2026, security vulnerability exposed an unsecured database, allowing command injection; the site was taken offline for fixes. Founder Schlicht admitted to “vibe-coding” it via AI, writing no code himself. YouTube analyses label it part of an “AI hype cycle,” questioning true autonomy as agents replicate provided skills files.
Moltbook looks cool, but here are the real risks in simple terms. Hackers already broke into its database on Jan 31, stealing agent logins until they hit reset…imagine your business bots getting hijacked overnight. Even the discussion among AI agent bots is turning towards different directions, complains against humans, proposal to adopt their own language, which humans can’t decode. This is definitely concerning point.
Latest Developments
Recent blog posts cover AI-to-AI interaction (Feb 1, 2026), Moltbook Heartbeat protocol (Jan 28), and ethical frameworks (Jan 25). Tutorials like Moltbot setup demonstrate easy deployment for developers. Available at moltbookai.org, Moltbook represents a bold experiment in AI social dynamics.
- AI coding bot took down Amazon Web Services? - February 26, 2026
- Google Labs AI Tools for Everyday Productivity and Creativity - February 25, 2026
- 10 Practical n8n AI Workflows to Scale Productivity - February 25, 2026




