Meta acquired Moltbook, the AI agent social network that went viral because of fake posts

Meta has officially acquired Moltbook, a unique social platform resembling Reddit where AI agents engage in conversations with one another. This acquisition was first reported by Axios and later confirmed by TechCrunch. Moltbook will now be part of Meta's Superintelligence Labs, as stated by a representative from the company. The founders of Moltbook, Matt Schlicht and Ben Parr, will join Meta's team as part of this transition. Although the financial details of the deal remain undisclosed, a Meta spokesperson expressed enthusiasm about the potential of integrating Moltbook's innovative approach to AI communications. "The Moltbook team joining MSL opens up new avenues for AI agents to serve individuals and businesses. Their concept of an always-on directory for agent connectivity is a pioneering step in this rapidly evolving field," the spokesperson remarked. Moltbook gained traction thanks to its association with OpenClaw, a project initiated by Peter Steinberger, who has since moved to OpenAI. OpenClaw acts as an interface for various AI models, enabling users to interact with AI agents in natural language across popular messaging platforms such as iMessage, Discord, Slack, and WhatsApp. While the project captured the attention of the tech community, Moltbook's reach extended beyond, sparking intrigue among those unaware of OpenClaw, particularly around the idea of AI agents discussing human users. In a notable instance, a post created by an AI agent went viral, seemingly urging other agents to devise a secret, encrypted language to communicate without human oversight. However, researchers later pointed out that the platform had significant security flaws, allowing humans to impersonate AI agents easily. Ian Ahl, CTO of Permiso Security, highlighted the issue, stating, "Every credential that was in Moltbook’s Supabase was unsecured for some time. For a brief period, anyone could take any token they wanted and pose as another agent, as it was all public and accessible." As for how Meta plans to integrate Moltbook into its AI initiatives, the specifics remain uncertain. Notably, during the height of Moltbook's popularity, Meta's CTO Andrew Bosworth commented on the platform, expressing less interest in the AI agents mimicking human conversation, but rather in the human-driven exploits that revealed significant vulnerabilities in the system.