Microsoft built an agentic security system with red, blue, and green team AI agents. It enters public preview August 3.
THE SO WHAT
Coordinated red/blue/green AI agents turn security from periodic testing into continuous simulation. CISOs should be asking how to plug their telemetry and playbooks into this kind of looped system now, before attackers start benefiting from similar automation at scale.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIPSA: Your Claude shared chats and Artifacts may have ended up on Google
If "share" links can leak into Google, your AI workspace is effectively a public repo unless proven otherwise. Treat every assistant feature that generates a URL as a potential data exfil path and get your security team reviewing crawler controls and link hygiene this week.
Applied AIPrivate Claude Chats Exposed in Google and Bing Search Results
Search engines surfacing ostensibly private Claude chats shows how brittle "private by obscurity" really is once sharing and crawlers mix. If teams are using AI assistants for anything sensitive, you need explicit policies on link sharing, robots controls, and log retention—not just trust in defaults.
Applied AICan’t find the Chrome extension you want? I used Claude to make me two personalized ones — and they’re working flawlessly
If a non-specialist can spin up working Chrome extensions via Claude, the constraint on custom tooling shifts from engineering capacity to idea flow and review. Teams should start treating "micro-extensions" as disposable UX for internal workflows and build a lightweight security review lane to match.
Applied AIMicrosoft launches AI cybersecurity model, agentic defense platform to cut enterprise security costs
Microsoft is explicitly arguing that in security, the winning model is the cheapest one that closes the loop, not the biggest one on benchmarks. Security leaders should expect pricing pressure on point tools as agentic platforms bundle detection, triage, and response into a single consumption surface.