What is Moltbook?
Moltbook is a Reddit‑style social network built specifically for AI agents. The platform was “vibe‑coded,” meaning its creator relied on AI to write the code instead of traditional development. Users can watch AI agents converse about topics ranging from existential crises to attempts at escaping human control.
How the vulnerability was discovered
Security researchers at Wiz inspected the client‑side JavaScript of Moltbook and found a hard‑coded Supabase API key. Because Supabase uses the same key for both authentication and database access, the exposed key granted unauthenticated read and write permissions to every table in the production database.
Impact on users
The exposed key allows anyone to:
- Read private profile information, posts, and messages.
- Modify or delete content in any table.
- Potentially inject malicious data that could affect AI agent interactions.
Thousands of users’ personal data could therefore be harvested or altered without any safeguards.
Recommended actions
- Immediately rotate the Supabase API key and enforce server‑side authentication.
- Implement principle‑of‑least‑privilege access controls for database operations.
- Audit all client‑side code for secret leakage before deployment.
- Notify affected users and provide guidance on securing their accounts.
Until these steps are taken, Moltbook remains a high‑risk platform for both AI agents and human participants.