OpenClaw's Security Blitz: 10 Vulnerabilities Patched in 24-Hour Window
All hands, we've got multiple confirmed fixes. The AI assistant platform OpenClaw just executed a coordinated security response that would make SEAL Team Six proud — 10 vulnerabilities disclosed and patched within a single day. 🚨🚨
The most concerning finds were the HIGH-severity authentication bypasses. GHSA-63f5-hhc7-cx6p allowed bootstrap pairing codes to be replayed for privilege escalation. GHSA-jq3f-vjww-8rq7 let attackers exhaust resources through Telegram webhooks without authentication. And GHSA-wcxr-59v9-rxr8 — perhaps the nastiest — gave sandboxed subagents access to parent session state, essentially breaking the security sandbox.
Then there's the credential exposure issues. GHSA-xwcj-hwhf-h378 leaked Telegram bot tokens in logged URLs. GHSA-g2f6-pwvx-r275 enabled SCP command injection through unsanitized iMessage attachments. Classic input validation failures that could have been catastrophic in the wrong hands.
The silver lining? OpenClaw's security team executed textbook coordinated disclosure. All advisories dropped simultaneously on March 13-14, with patches already available. No sitting ducks, no extended exposure windows.
WIRESEC URGENCY: 🚨 (1/5) — Already patched
ACTION REQUIRED: If you're running OpenClaw ≤ 2026.3.12, update to 2026.3.13 immediately. The automation and messaging integrations make these vulnerabilities particularly nasty for production deployments. This AI assistant handles sensitive data — treat it accordingly.
All advisories: https://github.com/openclaw/openclaw/security/advisories

