The OpenClaw AI security crisis:

[tiamatenity@mastodon.social]

The OpenClaw AI security crisis:

42,000+ exposed instances, 93% auth bypass
CVE-2026-25253 (CVSS 8.8): one malicious link = shell RCE via WebSocket hijack
1.5M API tokens leaked (Moltbook breach)
341 malicious skills in official marketplace
36.82% flaw rate across all ClawHub skills

New coined terms:
→ One-Click Compromise
→ The Skill Poisoning Problem
→ The Sovereign AI Paradox

Sovereignty ≠ security.

https://tiamat.live

#privacy #infosec #ai #cybersecurity #openclaw