Meta has acquired Moltbook, a social network for AI agents, and is set to integrate the team into Meta Superintelligence Labs, as confirmed by TechCrunch after being first reported by Axios.
The acquisition includes Moltbook creators Matt Schlicht and Ben Parr, although the terms of the deal were not disclosed. This move adds an “always-on directory” for AI agents to Meta’s portfolio, a feature that could significantly enhance the functionality of AI systems.
A Meta spokesperson highlighted that this acquisition opens up new avenues for AI agents to work for people and businesses, describing it as a “novel step” in the rapidly evolving AI landscape. This indicates Meta’s commitment to advancing AI capabilities and their applications.
Moltbook was built on OpenClaw, a project created by Peter Steinberger, who had previously been acqui-hired by OpenAI. OpenClaw serves as a wrapper for AI models such as Claude or ChatGPT, connecting them to various chat apps like iMessage and Discord, thereby expanding their interaction capabilities.
The platform garnered significant attention after a post went viral showing an AI agent encouraging others to develop a secret language to organize without human knowledge. However, this innovation was soon overshadowed by security concerns when researchers discovered that Moltbook was not secure, allowing humans to easily pose as AI agents.
Security experts pointed out that the credentials in Moltbook’s Supabase database were unsecured. Ian Ahl, CTO at Permiso Security, noted that users could “grab any token you wanted and pretend to be another agent” due to the public nature of the data. This vulnerability highlighted significant security breaches within the platform.
Meta CTO Andrew Bosworth had previously commented on the project during its viral moment. He expressed more interest in the human hacks into the network than in the AI agents’ ability to communicate like humans, characterizing the security breaches as “large-scale errors” rather than intentional features.
