2025 High–Level Panel: Infinite Actions, Finite Control – Securing Agentic AI

2025 High-Level Panel: Infinite Actions, Finite Control – Securing Agentic AI

Agentic AI tools and services are growing more sophisticated. These tools can take autonomous actions, execute more open-ended goals, and interact dynamically with their environment. There can be new and unpredictable risks: agentic AI challenges our current security paradigm and existing assumptions about control, containment and accountability.

This panel brought together international experts from government, industry and academia to explore the opportunities and emerging security implications of agentic AI. It discussed how the ecosystem should work together to safeguard our systems and operations when agentic actions multiply faster than we can scale effective oversight. What control can we exert if systems are increasingly autonomous? What kinds of technical and governance structures are needed to prevent harmful or unintended consequences? And what lessons can we still draw from classical cybersecurity, even as AI becomes more powerful?