# What is worth listening to before an AI proof of concept?

The **29 September 2026 #AuditTuesday** episode on rogue AI and identity governance discusses agents using legitimate user permissions. It is a YouAttest-hosted supplier perspective, not an independent product test; the episode description is useful preparation for questions about delegated access. ([buzzsprout.com](https://www.buzzsprout.com/2296327/episodes/19884818-worried-about-rogue-ai-start-with-identity-governance-audittuesday-w-justin-roy-and-jeff-kushner?t=0))

**Bring one question to the demonstration:** “If this person loses access now, what happens to every agent acting on their behalf?”