- Updated: February 2, 2026
- 2 min read
Moltbook Hack: AI Social Network Breach Exposes Millions of API Keys – AI Security Insights
Security researchers at Wiz uncovered a massive data leak affecting Moltbook, an emerging social network designed for AI agents. A misconfigured Supabase database left millions of records publicly accessible, revealing API keys, email addresses, private messages, and even write‑access credentials.
The exposed dataset included over 2 million API keys that could be used to interact with Moltbook’s backend services, effectively granting attackers full control over user accounts and the platform itself. The breach highlights a critical oversight in cloud database permissions and underscores the growing attack surface of AI‑driven applications.
Key takeaways for the AI community:
- Zero‑trust configuration: Cloud databases must enforce strict access controls and never expose credentials publicly.
- Continuous monitoring: Automated scans for open buckets or misconfigured services can catch leaks before they are exploited.
- Secret management: Store API keys and tokens in dedicated secret‑management solutions rather than hard‑coding them in code or configuration files.
For a deeper technical analysis, read the original Wiz investigation here. Our own coverage on related topics includes AI security best practices, effective data breach response, and risks in generative AI platforms.
Stay informed and protect your AI workloads – the Moltbook incident is a stark reminder that even cutting‑edge AI services are vulnerable when basic security hygiene is ignored.