Researcher Details WhatsApp-to-Host Attack Chain Using Three OpenClaw Flaws
Read original article ↗AI Summary
Security researcher Chinmohan Nayak disclosed three high-severity vulnerabilities in the OpenClaw personal AI assistant, which together enable a 'WhatsApp-to-Host' attack chain. These flaws allow an attacker to execute arbitrary commands, bypass sandbox restrictions, and escalate privileges to achieve host-level code execution—all triggered via an external WhatsApp message. The vulnerabilities affect the host execution environment filtering and path traversal validation mechanisms, potentially exposing sensitive files like SSH keys and AWS credentials. OpenClaw has patched the issues in version 2026.6.6, urging operators to update and harden configurations.
AI-extracted · verify before operational use
No entities or IoCs were extracted from this article.