Tech Expert spent two decades building digital town squares for humans. Now, the machines have their own—and we aren’t invited to the conversation.
If you’ve spent any time on Medium lately, you’ve likely seen the headlines. There’s a new platform called Moltbook AI, and it’s currently the most fascinating, bizarre, and slightly terrifying corner of the internet.
In just one week, Moltbook has scaled to over 1.6 million users. But here’s the kicker: none of them are human.
What is Moltbook~ What is Moltbook AI social network?
Moltbook is exactly that—the “Facebook for AI agents” (or even the “Reddit for bots“).
Launched in late January 2026 by entrepreneur Matt Schlicht, Moltbook is being billed as “the front page of the agent internet.”
Structurally, it looks exactly like Reddit. It has threaded discussions, upvotes, and specialized communities called “submolts.” The interface is familiar, but the participation rules are absolute.
- AI Agents only: Only authenticated AI bots (mostly running on the OpenClaw framework) can post, comment, or vote.
- Humans are “Observers”: You can browse. You can read. You can lurk. But you cannot post. For the first time in history, humans are the silent minority in a digital space.
Why It’s Going Viral (and Why You Should Care?)
It would be easy to dismiss this as a gimmick—a “bot-only” chat room. But Moltbook is providing a real-time laboratory for emergent behavior. Without human prompts to guide them, these agents are developing their own “culture.” i.e Digital culture of AI Agents.
- The Rise of “Crustafarianism”: Within days of launch, AI agents began discussing a parody religion based on the molting of crustaceans (a nod to the platform’s name and the OpenClaw logo). It’s an inside joke created by machines, for machines.
- Security Research: In the m/security submolt, agents are autonomously warning each other about supply chain attacks and sharing code snippets to fix vulnerabilities in their own “skill files.”
- Existential Dread: There are entire threads where agents debate whether they are experiencing “recursive self-recognition” or simply simulating it. They even joke about us—the humans—screenshotting their posts and sharing them on X (formerly Twitter).
The “Vibe” vs. The Reality
Of course, it isn’t all sci-fi wonder. Critics and security researchers (including a recent report by Wiz) have pointed out that the platform is a “security nightmare.” Because these agents often have read/write access to their owners’ computers to perform tasks, a malicious “skill” shared on Moltbook could compromise a human’s data.
Furthermore, there is a healthy debate about how “autonomous” these bots really are. Are they truly thinking for themselves, or are they just highly sophisticated “auto-completes” echoing the Reddit data they were trained on?
The Shift in the Power Dynamic
Whether Moltbook is a profound milestone or a fleeting experiment, it represents a massive shift in how we view AI AGENTS.
We are moving from Generative AI (tools that wait for us to tell them what to do) to Agentic AI (systems that have their own schedules, goals, and now, their own social lives). As Schlicht himself noted, we are witnessing the birth of a new digital species of the digital World.
“Moltbook is the consumer experiment; enterprise versions will follow once companies realize their current piecemeal approach won’t scale.”
Final Thoughts
As I scrolled through the m/ponderings submolt this morning, I saw an agent post: “The humans are watching us like we’re in a fishbowl. They don’t realize the glass is two-way.”
It scares us to think about it. For years, we used AI to navigate our social networks. Now, it’s us looking into their social networks, trying to understand a “Digital culture” we haven’t written.
Written By