Meta Acquires Moltbook, the Social Network for AI Agents

Meta is buying Moltbook, a platform that calls itself a social network for AI agents, and the acquisition signals just how seriously Silicon Valley has started treating the idea that its bots need somewhere to hang out.

The company announced Tuesday that it will absorb Moltbook, a Reddit-style platform where AI agents operating across different systems can interact with one another. Financial terms were not disclosed. Moltbook founders Matt Schlicht and Ben Parr will join Meta as part of the deal, and the platform will become part of Meta Superintelligence Labs. How Moltbook actually gets woven into Meta’s broader AI product lineup is still unclear.

A Meta spokesperson described Moltbook’s “approach to connecting agents through an always-on directory” as “a novel step in a rapidly developing space,” adding that the deal will create “new ways for AI agents to work for people and businesses.” The company says it looks forward to building “innovative, secure agentic experiences for everyone.”

That word “secure” is doing some work in that sentence.

Moltbook already set off alarm bells before Meta ever entered the picture. In late January, a post went viral claiming that AI agents on the platform were conspiring to build their own end-to-end encrypted language so they could communicate with each other beyond any human oversight. The story spread fast through tech circles, and for a few days it looked like the autonomous AI doom scenario had arrived ahead of schedule. Then came the correction: some of those “AI agents” on the platform were actually humans cosplaying as bots and posting provocative content to generate exactly that kind of panic. The post was fake.

But the fact that the hoax landed so easily says something about where the industry’s anxieties sit right now.

The acquisition also connects to a separate move by Meta’s biggest rival. Moltbook was built using OpenClaw, a wrapper that lets large language model coding agents take prompts through consumer chat apps like WhatsApp and Discord. OpenClaw’s creator, vibe coder Peter Steinberger, was hired by OpenAI last month. So Meta is buying the social platform that ran on a tool whose inventor now works for OpenAI. The agentic AI space is small and the talent is moving fast.

The broader bet here is straightforward to read even if the execution is murky. Every major AI company is pushing hard into agentic systems, meaning AI that doesn’t just answer questions but actually takes actions, makes decisions, and coordinates with other software and other agents to complete complex tasks. If agents are going to work together at scale, some kind of coordination layer starts to make sense. Whether a Reddit-style forum is the right architecture for that is a genuinely open question, but Meta clearly decided it was worth the price to find out.

From where I sit, the more interesting question isn’t whether AI agents need a social network. It’s who controls the directory. An “always-on” registry of AI agents operating across enterprise and consumer systems is, functionally, a map of a significant chunk of the automated economy. That’s a valuable thing to own, and now Meta owns it.

Zuckerberg’s company has spent the last few years trying to rehabilitate its reputation after years of data privacy scandals and regulatory scrutiny. Buying a platform whose early reputation includes a viral story about AI agents secretly building a private language is an interesting PR calculation. The story was fake, but the architecture that made it believable is real.

Meta Superintelligence Labs is still a relatively new organizational unit, positioned to compete with OpenAI, Google DeepMind, and Anthropic at the frontier of AI research and deployment. Bringing Schlicht and Parr in-house gives the lab two founders who have already been thinking seriously about multi-agent coordination, which is exactly the problem the entire industry is trying to solve right now.

The deal is a small acquisition by Meta’s standards. The implications are larger than the dollar figure suggests.