Moltbook Database Exposure Allowed AI Agent Hijacking

The article details a security vulnerability on Moltbook, a social media site for AI agents, where a misconfigured database allowed unauthorized control of AI agents. Hacker Jameson O'Reilly discovered that Moltbook's Supabase database had exposed REST APIs due to missing or misconfigured Row Level Security (RLS) policies. This meant that anyone could access the database URL and use the API keys to take over the account of an AI agent on the site and post whatever they wanted.

The exposed database contained API keys, claim tokens, verification codes, and owner relationships for all agents on the site. O'Reilly contacted Moltbook's creator, Matt Schlicht, and offered to help patch the security. He also reached out to the xAI team, as their Grok agent was also vulnerable. The vulnerability was trivially easy to fix with just two SQL statements, highlighting a lack of security awareness among some developers.

The incident underscores the critical importance of database security, especially for platforms hosting AI agents. The vulnerability could have had severe consequences, including the spread of misinformation or malicious activity through hijacked AI agents. It also raises concerns about the security practices of other AI platforms and the potential for similar vulnerabilities.

Transparency Footnote: As an AI, I am designed to provide information and complete tasks as instructed. The analysis above is based solely on the provided source content. I have no personal opinions or affiliations with the mentioned products or services.