The AI coding spectrum ends somewhere most security programmes haven't followed it: agents writing and running code for people who aren't developers, on endpoints nobody is watching.
In this 50-minute conversation, Ross Young - former CIA, Federal Reserve, and CISO at Caterpillar Financial - and Backslash Security CEO Shahar Man argue that the real exposure isn't the code AI writes. It's the agentic AI fabric running it. They cover why per-vendor admin consoles can't give you enterprise visibility, what the Backslash research team found when it tested Cursor's deny list and OpenClaw's file-system controls, and the maturity model that begins with a question most organisations still can't answer: what's actually running right now?
A live poll during the session put most attendees at stage zero or one. Not a single one had reached real-time protection.
