Researchers Reveal Six New OpenClaw Vulnerabilities

Endor Labs disclosed six vulnerabilities in the OpenClaw agentic AI assistant — including high-severity SSRF, missing webhook authentication, and path traversal issues (CVE-2026-26322, CVE-2026-26319, CVE-2026-26329 and multiple GHSA advisories) — noting public exploit code for several CVEs, tens of thousands of exposed misconfigured instances, and active targeting by infostealer malware, creating significant enterprise exposure.