The identity-based access model that has anchored zero trust for the past decade assumes a human is behind every request, but agentic AI breaks that assumption entirely. A single coding agent can fan out across models, MCP servers, APIs, and data stores in seconds, pursuing a goal with no judgment about consequence and no pause for human review.
Join Matt Bromiley, SANS Certified Instructor, on December 10, 2026 at 10:30AM ET in this SANS Security Lab where they will walk through how Cisco's SASE platform extends zero trust from access control to action control.
Through expert discussion and live demonstration, they'll break down key challenges facing security teams, including:
- agent sprawl across a fragmented tool ecosystem
- inconsistent policy enforcement across MCP servers and APIs
- controlling non-deterministic agent behavior before it turns into unauthorized action
Attendees will see live product demos covering agent discovery and identity mapping, gateway-level policy enforcement, and observation-mode deployment for tuning controls without disrupting workflows, plus a practitioner readiness self-assessment to gauge where their organization stands today.
