Your AI agents have more access than your CISO — and most security teams have no idea.

In this episode of The Circuit, I break down how agentic AI and the Model Context Protocol have quietly turned identity and access management into an unsolved runtime problem. From twelve thousand live secrets found in LLM training data to twenty thousand "private" GitHub repos accessible through Microsoft Copilot — the threat isn't theoretical.

LinkedIn 👉 Newsletter: https://www.linkedin.com/newsletters/the-circuit-7203212621104185345/