The new Moltbook is dealing with a serious vulnerability. Security researchers found a misconfigured Supabase database that allowed anyone to access all data, including 1.5 million API tokens and 35,000 email addresses. The data leak was fixed within a few hours.
Moltbook launched a few days ago as a social platform where AI agents post and communicate with each other. People are only allowed to watch. The platform caught the attention of several well-known names in the tech industry.
The founder of Moltbook said he had completely vibe-coded the platform, a term for quickly building applications without writing code yourself. However, according to the creator, this led to dangerous security oversights, similar to previous vulnerabilities at DeepSeek and Base44.
Wiz security researchers found a Supabase API key in the website’s client-side JavaScript within minutes. This key gave unauthenticated access to the entire production database, including read and write permissions on all tables. This is a public API key that is normally secure when Row Level Security (RLS) is properly configured.
However, without an RLS policy, this key grants full database access to anyone who has the key. Moltbook lacked this crucial line of defense.
People behind the AI network
The ratio of agents to humans was 88:1. Anyone could register millions of agents through a simple loop. People could also post content as if it came from AI agents by sending a basic POST request. The platform had no mechanism to verify whether an “agent” was actually AI or simply a human with a script.
The exposed data included 1.5 million API authentication tokens, 35,000 email addresses, and private messages between agents. The researchers immediately reported the issue to the Moltbook team, who fixed it within hours. All data collected during the investigation has been deleted.