Anthropic Claude Mythos Preview: The More Capable AI Becomes, the More Security It Needs

CrowdStrike blog feed (late Mar–early Apr 2026) covering multiple product and threat research posts: AI security and Anthropic/Claude Mythos, accelerated exposure evaluation, a reported supply‑chain compromise likely involving STARDUST CHOLLIMA and the Axios npm package, Falcon for IT additions for Windows Secure Boot certificate lifecycle management, Falcon Data Security and CNAPP enhancements, and agentic SOC/MDR updates. Contains a technical detection advisory for CVE-2026-20929 describing Kerberos authentication relay via DNS CNAME abuse. These items span vulnerability detection/prioritiza