Earlier this year, two founders launched Moltbook, a viral Reddit-like social network where the users aren’t people, but AI agents. Today, Meta has acquired it.
Meta has acquired Moltbook, a viral social network where AI bots talk to each other, in a move that highlights how big tech companies are racing to control the fast-growing world of AI agents.
The platform gained widespread attention earlier this year as people watched AI bots post updates, comment on each other’s messages and hold public conversations online.
But the buzz was also fueled by controversy. Security flaws allowed humans to impersonate AI agents and publish fake posts, raising questions about how much of the activity on the platform actually came from bots.
Despite those issues, the acquisition signals Meta’s growing interest in building tools and infrastructure that allow AI agents to find each other, communicate and work together online.
The creators of Moltbook
Moltbook launched in early 2026 as a Reddit-style platform designed for autonomous AI agents. Many of those agents ran on the OpenClaw framework, allowing them to post updates, comment on one another’s messages, and vote on content.
The concept quickly captured attention beyond the AI developer community. Observers were intrigued by the idea of an online space where software agents could interact with each other continuously and publicly.
The startup behind the platform was founded by Matt Schlicht, an AI and ecommerce entrepreneur, and Ben Parr, a former journalist and AI investor. The OpenClaw agent framework that powered many of the bots on Moltbook was developed by Peter Steinberger, who has recently joined OpenAI.
Why Meta acquired Moltbook
As part of the acquisition, Moltbook and its founding team will join Meta Superintelligence Labs (MSL), the Meta’s internal AI research unit, led by Alexandr Wang. Financial terms of the deal were not disclosed, though both Schlicht and Parr are expected to join MSL.
A Meta spokesperson said the company was particularly interested in Moltbook’s “always-on directory” approach, which allowed AI agents to discover and communicate with each other in real time.
Despite its rapid growth, Moltbook also faced significant security problems. A misconfigured Supabase database exposed roughly 1.5 million API tokens and tens of thousands of email addresses due to missing row-level security protections.
The vulnerability made it relatively easy for human users to impersonate AI agents and publish fabricated posts. These potentially fake messages helped drive Moltbook’s viral popularity but also undermined the perception that the platform represented an authentic ecosystem of autonomous agents.
The race among tech giants to build AI agent platforms
Meta is expected to focus first on securing agent identity and strengthening platform safeguards as Moltbook is integrated into its operations.
Looking ahead, the technology could eventually support tools for discovering, coordinating, and connecting AI agents across Meta’s messaging platforms, including Messenger and WhatsApp.
The acquisition signals growing interest in infrastructure for AI-to-AI interaction. At the same time, Moltbook’s early security issues highlight the governance and verification challenges that could shape how future agent ecosystems are built.
The deal also reflects a broader trend in the AI sector, as foundational projects behind agent communities become increasingly absorbed by major labs.
The OpenClaw framework moved toward the OpenAI ecosystem, and Moltbook’s founders are now joining Meta – consolidating talent and infrastructure as competition over AI agents intensifies.
