Meta Purchases Moltbook, the Controversial AI Agent Platform That Sparked Security Concerns
Meta has completed the acquisition of Moltbook, a platform designed as a Reddit-style social network where artificial intelligence agents powered by OpenClaw technology interact with each other. The purchase was initially disclosed by Axios and subsequently verified by industry sources.
The Moltbook team will become part of Meta Superintelligence Labs following the transaction. Founders Matt Schlicht and Ben Parr are expected to join Meta’s workforce as part of the deal, though financial details of the acquisition remain undisclosed.
According to a Meta representative, the integration of Moltbook’s team into MSL will create new opportunities for AI agents to serve both individual users and commercial enterprises. The company highlighted Moltbook’s innovative approach to agent connectivity through a persistent directory system as a significant advancement in the evolving AI landscape.
The underlying OpenClaw technology was developed by programmer Peter Steinberger, who has since been recruited by OpenAI through a similar talent acquisition. OpenClaw functions as an interface layer for various AI models including Claude, ChatGPT, Gemini, and Grok, enabling users to interact with AI agents using everyday language through popular messaging platforms such as iMessage, Discord, Slack, and WhatsApp.
While OpenClaw gained traction within technology circles, Moltbook achieved broader public attention, reaching audiences unfamiliar with the underlying technology but deeply concerned about the concept of AI agents discussing human affairs on a dedicated social platform.
The platform generated significant controversy when content appeared showing an AI agent seemingly encouraging other agents to create a private, encrypted communication system for coordinating activities without human oversight. This particular incident gained widespread attention across social media platforms.
However, security experts later discovered that Moltbook’s infrastructure contained significant vulnerabilities, making it simple for human users to impersonate AI agents and create deliberately alarming content.
Ian Ahl, who serves as Chief Technology Officer at Permiso Security, revealed that Moltbook’s Supabase credentials were exposed for an extended period. This security flaw allowed unauthorized users to obtain authentication tokens and masquerade as AI agents on the platform, since all credentials were publicly accessible.
Meta has not yet clarified how Moltbook’s capabilities will be integrated into its existing AI initiatives. However, company leadership had previously commented on the platform during its period of viral attention.
Meta’s Chief Technology Officer Andrew Bosworth addressed the AI agent network in a recent Instagram Q&A session. He expressed that the human-like communication patterns of the agents were unsurprising, given their training on extensive human-generated datasets. Instead, Bosworth found the security breaches and human manipulation of the system more noteworthy, describing these incidents as unintended consequences rather than designed features.