On the latest Security Visionaries podcast, host Emily Wearmouth invites Steve Riley back to demystify another key acronym in the AI world: MCP, or model context protocol.  They break down what MCP actually is, how it functions, and why it is critical to understanding AI interactions. From there, they pivot into a discussion about the critical security implications of MCP, covering the risks of unauthenticated servers and the necessity of building both an authorization layer and an external policy layer to ensure granular access, data security, and compliance as AI agents proliferate. As always, this conversation with Steve offers illuminating context and perspective on one of the key security terms going into 2026. You won’t want to miss it.