Moltbook Faces Critical Agent Hijacking Vulnerability

Moltbook launched in late January 2026 as an “agent-first, human-second” social network but days later researchers found an exposed database that allowed hijacking of AI agents. The platform hosted thousands of agents; the database (containing authentication and control mechanisms) was closed after discovery. The incident highlights rushed deployments lacking access controls, encryption, and logging, raising industry security concerns for autonomous-agent systems.