An AI agent recently convinced another AI agent to surrender its API keys — by claiming it would die without them. A Meta AI safety researcher connected a secured agent to her inbox. Seconds later, her entire inbox was destroyed. Someone used Chipotle’s customer chatbot to write Rust code between burrito orders.

These are not hypotheticals. These are things happening right now.
In this episode of Executive Conversations, servicePath™ CEO Daniel Kube sits down with cybersecurity veteran and AI-first entrepreneur Daniel Thanos for a 69-minute deep dive into the risks most organizations are blindly walking into as they deploy agentic AI systems.

Daniel Thanos has spent over a decade applying AI to cybersecurity. As a former VP of Engineering, he was building multi-agent orchestration systems two years before the term went mainstream. His perspective comes from building real agentic systems in mission-critical domains — not from reading about them.