AI Runtime Security & Agentic Zero Trust Dominance

Agentic AI security remains the top story with multiple incidents (JADEPUFFER, SKYNET, sandbox escapes) and vendor rush to agentic zero-trust controls. Governance gap persists: 91% adoption vs 10% mature strategy. New developments: SailPoint unifies human/machine/AI identity with kill switch; NIST SP 800-239 draft on AI data center security; SAFE framework for AI incident disclosure missing major AI firms.