Active attack: Dirty Frag Linux vulnerability expands post-compromise risk

The feed highlights multiple active and emerging threats: a newly disclosed Linux local privilege escalation dubbed “Dirty Frag” affecting kernel networking/memory-fragment handling that enables reliable escalation from unprivileged users to root (limited in‑the‑wild activity and Microsoft Defender detections reported); research showing prompt-injection vulnerabilities in AI agent frameworks that can lead to remote code execution; and a confirmed high-severity Linux vulnerability (CVE-2026-31431, “Copy Fail”) with an exploit observed in the wild that enables root escalation across cloud and K8