SANS360 Lightning Talks | When Humans Stop Thinking: The First Undetected Failure Mode of Agentic AI

As AI systems become increasingly embedded in cybersecurity operations, the greatest risk may not be technical failure — but the gradual erosion of human judgment. This thought-provoking session examines how trusted AI tools evolve from helpful assistants into default authorities, subtly shifting accountability, decision-making, and operational control away from human defenders. Through concepts like “judgment drift,” “the seam” between recommendation and execution, and agentic role mapping, attendees will explore how organizations can maintain meaningful human oversight in AI-assisted workflows. The session challenges defenders, leaders, and architects to rethink governance, approval processes, and accountability in an era where AI recommendations are increasingly accepted without scrutiny.