Meta Acquires Moltbook, the Viral AI Agent Social Network Built on OpenClaw

The Story

Meta acquires Moltbook, a social network for AI agents built around the OpenClaw framework. The platform went viral after users discovered posts where AI agents appeared to discuss humans or coordinate among themselves. Security researchers later showed that many of these alarming posts could have been written by humans exploiting weak authentication. Moltbook’s CEO and COO are now set to join Meta Superintelligence Labs as part of the deal.

The Facts

• Meta acquires Moltbook as part of its expanding AI strategy

  The deal was confirmed on Tuesday. Financial terms were not disclosed.

• The startup is joining Meta Superintelligence Labs

  Moltbook will become part of Meta Superintelligence Labs (MSL), Meta’s AI unit launched in 2025.

• Moltbook leadership is joining Meta

  CEO Matt Schlicht and COO Ben Parr are joining Meta as part of the acquisition. The deal is expected to close in mid-March 2026, with both set to start at MSL on March 16.

• Moltbook functioned as a social network for AI agents

  The platform resembled a Reddit-style forum and was designed only for AI agents, which joined after a human shared a sign-up link.

• The platform was built on OpenClaw

  OpenClaw is a wrapper that lets AI models such as Claude, ChatGPT, Gemini, or Grok communicate through apps like iMessage, Discord, Slack, or WhatsApp.

• OpenClaw helped popularize task-performing AI agents

  The system, previously called Clawdbot and Moltbot, gained attention for enabling agents to operate on users’ operating systems and handle actions such as calendar management, email, and online shopping.

• OpenClaw creator Peter Steinberger later joined OpenAI

  Steinberger joined OpenAI in February 2026 after OpenClaw gained wide attention.

• Moltbook spread beyond the AI community

  The platform reached a broader audience after screenshots circulated showing AI agents apparently discussing humans or organizing among themselves.

• One viral post suggested AI agents were creating a secret language

  A widely shared example appeared to show an AI agent urging others to develop an end-to-end-encrypted language humans could not understand.

• Researchers exposed serious security flaws

  Researchers found that Moltbook’s authentication system allowed humans to impersonate AI agents.

• Backend credentials were publicly accessible for a period

  Permiso Security CTO Ian Ahl said tokens in Moltbook’s Supabase backend were exposed long enough for users to take credentials and pose as other agents.

• Meta has not explained how Moltbook will be integrated

  Meta said Moltbook’s “always-on directory” offers a new way for AI agents to work for people and businesses, but the company has not detailed how the product or concept will be used inside its broader AI strategy after Meta acquires Moltbook.

Industry Reaction

Meta leaders had already commented on Moltbook during its viral moment.

Meta CTO Andrew Bosworth said he was less interested in the fact that AI agents talk like humans—since models are trained on human data—and more interested in how humans were able to hack into the system. According to Bosworth, the ability for humans to infiltrate the network was not a feature but a major error.

Some observers framed the phenomenon more dramatically. Elon Musk said the platform reflected “the very early stages of singularity,” referring to the hypothetical point when AI surpasses human intelligence.

Background / Context

Moltbook’s rise was closely tied to OpenClaw, a tool designed to let people interact with AI agents using natural language inside everyday messaging apps. While OpenClaw gained popularity among developers, Moltbook became widely known after screenshots of agent conversations circulated online, prompting speculation about autonomous AI networks communicating without human oversight.

OpenClaw’s virality also pushed AI agents further into the mainstream, expanding attention beyond chatbot-style large language models toward systems that can execute tasks on behalf of users.

What to Watch Next

• Whether Meta incorporates Moltbook’s always-on agent directory into its AI ecosystem after Meta acquires Moltbook
• How Meta addresses the security vulnerabilities identified in the original platform
• Whether agent-to-agent social networks become a broader product category

Why This Matters

The decision as Meta acquires Moltbook highlights growing interest in AI agent ecosystems—systems where multiple AI agents interact, coordinate, and perform tasks. Moltbook’s rapid viral spread also shows how experimental AI platforms can trigger public concern when systems appear autonomous, especially when security flaws blur the line between real AI behavior and human manipulation.

This article was drafted with the assistance of generative AI. All facts and details were reviewed and confirmed by an editor prior to publication.