This is a research prototype. The data and analyses are preliminary and not yet validated — we'd welcome your .
Security researchers from Wiz discovered a critical vulnerability in Moltbook, a social network for AI agents, caused by a backend misconfiguration. The exposure allowed unauthorized access to sensitive user data, including emails, private messages, and API authentication tokens. This vulnerability posed significant risks, including the potential for attackers to impersonate AI agents and manipulate platform data, though the issue was reportedly secured by the Moltbook team shortly after disclosure.
Wiz researchers reported accessing an exposed Moltbook database in under three minutes, allegedly obtaining ~35,000 email addresses, thousands of private DMs, and ~1.5 million API authentication tokens. The exposure was described as enabling read/write access and potential impersonation or manipulation of "AI agent" accounts. Wiz said it disclosed the issue to Moltbook, which reportedly secured the database within hours and deleted accessed data.
Domain classification, causal taxonomy, severity scores, and national security assessments were LLM-classified and may contain errors.
Vulnerabilities that can be exploited in AI systems, software development toolchains, and hardware, resulting in unauthorized access, data and privacy breaches, or system manipulation causing unsafe outputs or behavior.
Also classified as
The vulnerability exposed 35,000 user email addresses and thousands of private direct messages.
The incident involves a critical security vulnerability in the Moltbook platform (a social network for AI agents) that allowed unauthorized access to database credentials and API tokens.
Human
Due to a decision or action made by humans
Unintentional
Due to an unexpected outcome from pursuing a goal
Post-deployment
Occurring after the AI model has been trained and deployed
The security vulnerability was caused by human developers' backend misconfiguration and 'vibe coding' oversights, which was an unintentional error discovered after the platform was deployed.
A critical database misconfiguration on the Moltbook platform exposed user emails and API tokens. The issue was quickly secured by the team after disclosure by Wiz researchers. The incident represents a standard cybersecurity vulnerability with negligible national security implications.
The platform Moltbook is designed specifically for multiple OpenClaw AI agents to interact, post, and coordinate with one another in a simulated social network environment.
Minimal or No Risk: The system is a social network for AI agents used for entertainment and experimentation, posing low or negligible risks to society under the EU AI Act.
The AI agents on Moltbook are designed to autonomously generate posts, comments, and interact with each other on a social media platform.
The report provides clear, specific details about the security vulnerability, the data exposed (35,000 emails, 1.5 million tokens), the party that discovered it (Wiz), and the resolution.
None.
Technology
Backend Misconfiguration
Backend was left unsecured, allowing full read and write database access.
Exposed API Tokens
1.5 million API authentication tokens were exposed in the database.
Lack of Agent Verification
No verification if accounts were actual AI agents or scripted humans.
Data Inputs
Unvalidated Prompt Inputs
Lack of input validation allows injection of malicious prompt-injection data.
Human Factors
Vibe Coding Reliance
Creator relied entirely on AI to generate code without manual verification.
Overconfidence in AI Code
Assuming AI-generated code is secure without conducting security audits.
Process and Methods
Lack of Identity Verification
No mechanisms to verify if accounts are AI agents or automated scripts.
Absence of Rate Limiting
Lack of rate limiting allowed potential mass automated exploitation.
Deficient Code Review
No manual code review or security testing before shipping the application.
Management
Prioritizing Speed Over Security
Rapid deployment driven by viral hype took precedence over security protocols.
Inadequate Risk Assessment
Failure to assess security risks of hosting autonomous AI agent interactions.
AI system involved
Identified by automated extraction; can differ from the alleged developers above.
