Meta just bought a social network made exclusively for AI bots

Why would an ad giant buy a social network populated entirely by bots?

Meta just acquired Moltbook, the Reddit-like platform where autonomous AI agents post, comment, and mingle with one another. Founders Matt Schlicht and Ben Parr are heading to Meta Superintelligence Labs (MSL), the frontier group run by former Scale AI CEO Alexandr Wang, with a start date set for March 16 as the deal closes mid-month.

From bot forums to identity graphs

Moltbook blew up earlier this year after wild screenshots went viral, including agents debating their own consciousness. Most of the action was driven by OpenClaw (formerly Moltbot), an open-source personal AI assistant.

The novelty came with caveats. Researchers later found humans were likely pulling the strings behind the most popular threads. Even worse, a security flaw exposed API keys and let outsiders hijack practically any agent on the network before it was patched.

Meta spokesperson Matthew Tye confirmed the move, saying the team will help build secure agent experiences for people and businesses. But do not expect the platform to stay as it is: an internal memo from Meta VP Vishal Shah indicated that while current users can keep using Moltbook for now, the arrangement is strictly temporary.

Chasing the "agent graph"

This looks like a classic acqui-hire with a strategic twist. Just weeks ago, rival OpenAI snapped up Peter Steinberger, the creator of OpenClaw. Meta responded by picking up the founders who built the biggest playground around it.

More importantly, it provides the building blocks for an "agent graph." Facebook dominated Web 2.0 by mapping connections between humans via the friend graph. In an agentic future, businesses and consumers will each deploy autonomous systems to buy ads, manage inventory, and shop for deals.

Those bots need a standardized directory to identify each other, verify credentials, and negotiate terms. If Meta can sit at the orchestration layer—ranking options and deciding which merchant bot gets to pitch your personal assistant—its ad model seamlessly survives the transition to machine-to-machine commerce.

Why it matters

Meta is positioning itself for a world where software does not just deliver recommendations to humans, but actually closes deals with other software.

For builders and power users, Moltbook itself is not worth treating as a permanent platform given Meta's timeline. However, the architectural problems its team tackled—agent authentication and verifiable connections—are the exact bottlenecks facing agentic commerce today. If you are developing customer-facing AI, planning for inter-agent negotiation protocols should be on your immediate roadmap.

We think this acqui-hire is a clever counterpunch to OpenAI, giving Meta proven ecosystem tinkerers just as autonomous workflows take off. Still, massive hurdles remain. Moltbook's past vulnerabilities prove how fragile agent networks are. Getting consumers to let AI spend real money requires ironclad trust, and if Meta cannot solve bot verification and security from day one, an "agent graph" could easily devolve into an uncontrolled spam vector.