When anyone can control your AI agent
Without proper authentication, anyone who discovers your OpenClaw endpoint can send commands, access data, and abuse your AI agent's capabilities.
Unauthorized access occurs when individuals gain access to your OpenClaw without proper authentication. This is one of the most common and dangerous vulnerabilities in self-hosted AI deployments.
Many developers expose their OpenClaw on a public IP or domain for convenience, often with minimal or no authentication. This creates an open door for attackers who scan the internet for exposed AI endpoints.
Once an attacker gains access, they have full control over your AI agent—they can execute commands, access integrated services, consume your API credits, and potentially pivot to other systems on your network.
The most common issue—the OpenClaw endpoint is simply exposed without any authentication requirement.
Using default, common, or easily guessable passwords that attackers can brute-force.
API keys or session tokens exposed in logs, URLs, or client-side code.
Stealing valid session tokens through network interception or XSS attacks.
Using leaked credentials from other breaches to attempt login.
Security researchers regularly find exposed AI agent endpoints by scanning common ports and looking for telltale responses. In one case, a company's internal AI assistant was found exposed on the public internet with no authentication.
Attackers used the agent to: - Query internal databases for customer information - Send emails on behalf of employees - Access internal documentation and credentials - Generate content using the company's API credits
The breach wasn't discovered for weeks because there was no monitoring or audit logging in place.
When you self-host your OpenClaw, you're responsible for addressing these risks:
Clawctl includes built-in protection against unauthorized access:
Every request must be authenticated through our secure gateway. No anonymous access is possible.
Secure, rotatable API keys with granular permissions. Keys are never exposed in logs or error messages.
All traffic is encrypted in transit with modern TLS. Certificates are automatically managed and renewed.
Built-in rate limiting prevents brute-force attacks and abuse. Configurable limits per API key.
Every authentication attempt is logged. Failed attempts trigger alerts for potential attacks.
Whether you use Clawctl or not, follow these best practices:
Clawctl includes enterprise-grade protection against this threat and many others. Deploy your OpenClaw securely in 60 seconds.