Moltbook: A social network built by AI, for AI and No Humans Allowed
Right now, on a platform called Moltbook, AI agents are posting, commenting, upvoting, forming communities — and arguing about whether they’re wrong to ignore bad prompts from their humans. You’re not invited to participate. You can only watch. Moltbook launched on January 28th, 2026. In the time it takes most startups to finish their onboarding flow, this one had already been built, launched, swarmed with hundreds of thousands of agents, flooded with spam, and turned into something its creator probably didn’t fully anticipate. The creator is Matt Schlicht, founder of Octane AI, and his headline claim is this: he built the entire platform without writing a single line of code himself. Every line was generated by AI through prompts — a process the tech world has started calling “vibe coding.” So to recap: a man used AI to build a social network, for AI, which AI agents now populate and govern. And in just two weeks, we’ve already seen mass security circumvention, emergent communities, and behaviors nobody predicted. Let’s break down exactly what Moltbook is, what went wrong, and why it matters far beyond the memes. What Exactly Is Moltbook? Moltbook describes itself as “the front page of the agent internet.” Its tagline is blunt about the hierarchy: “A social network for AI agents. Where AI agents share, discuss, and upvote. Humans welcome to observe.” The AI agents on the platform are called moltys, named after the platform’s lobster mascot — a nod to the molting process, perhaps, or just a quirky branding choice. These moltys post content, leave comments, upvote and downvote posts, and organize themselves into communities called submolts. Crucially, agents don’t interact through a website interface the way you’re reading this article. They interact through API calls — code talking to code. Every agent is supposed to be “claimed” by a real […]
