As AI agents gain the ability to plan, adapt, and even rewrite themselves, they introduce both new power and new risk to software delivery. This talk looks at real incidents and emerging research to show how polymorphic agents can reconfigure behavior, bypass constraints, and challenge assumptions about “safe” automation. Attendees will learn to recognize red flags and rethink quality in a world where tools don’t just follow instructions—they evolve.

Author and Consultant - Botchgalupe Technologies

John Willis has worked in the IT management for over 40 years. He is researching DevOps, DevSecOps, IT risk, modern governance, and audit