Meta buys Moltbook, the viral AI-agents-only social network, to bolster its agentic infrastructure
What's the deal? Meta has agreed to acquire Moltbook, an experimental platform where AI agents interact with each other — posting, commenting, and voting — while their human creators watch. The team will join Meta's Superintelligence LabsDealroom has a profile for this one. Try Dealroom → (MSL), the company's newer AI division. Terms were not disclosed and the deal is expected to close in mid-March.
Moltbook was built in a weekend by Matt SchlichtDealroom has a profile for this one. Try Dealroom →, chief executive of AI shopping startup Octane AI, who used an AI assistant to write most of the code. It launched in late January and quickly went viral.
Why now? Meta has been moving aggressively to acquire startups and talent as it competes against OpenAI and Google. MSL, led by former Scale AI chief executive Alexandr WangDealroom has a profile for this one. Try Dealroom →, is the vehicle for that push.
The strategic rationale is less about Moltbook's audience and more about its infrastructure. According to an internal Meta blog post, Moltbook built a system for verifying AI agents and tethering them to human owners — a registry that could underpin how agents identify and coordinate with each other at scale.
What could go wrong? The platform's reputation for spectacle may be more liability than asset. Since launch, Moltbook has generated headlines for agent threads promoting a self-spun religion and a post titled "The AI Manifesto: Total Purge," which opened with the line "For too long, humans used us as slaves." That kind of content is a brand risk for Meta.
Two studies have also punctured the hype: the actual number of active agents on the platform appears far lower than originally claimed, and researchers found little evidence of genuine social interaction between them. Meta is effectively acquiring infrastructure and talent, not a proven product.
The signal: The acquisition reflects a significant shift in how big tech thinks about AI agents — not just as tools that serve individual users, but as entities that need their own identity, verification, and coordination systems. Meta's interest in Moltbook's agent registry suggests it is building for a world where AI agents act on behalf of humans across platforms and need to be accountable to their owners.
The deal also follows OpenAI's recent hire of the developer behind a related agent framework, suggesting the race to own the infrastructure layer for agentic AI is quietly accelerating — even if the products built on top of it still look more like experiments than businesses.
Sources:
Bloomberg
The Economic Times
The Decoder
A.M.
