OpenClaw SSH Vulnerability Enables Sandbox Escape and Arbitrary File Access

This critical vulnerability in OpenClaw, a framework for deploying AI agents, allows attackers to escape sandbox restrictions and access arbitrary files on both local and remote systems. The flaw can be exploited through prompt injection, meaning adversaries can manipulate AI agent interactions to trigger malicious file operations. With a CVSS score of 8.8, this represents a severe risk to any organization using OpenClaw for autonomous agent deployment.