AI Agents Expand Attack Surface at Unprecedented Scale

Atos' Stefan Mindermann on Supply Chain Risks and Autonomous Agents Rahul Neel Mani (@rneelmani) • December 16, 2025

Agentic artificial intelligence capabilities are accelerating enterprise transformation, but it is also driving a rapid and often underestimated expansion of the attack surface.

See Also: Going Beyond the Copilot Pilot - A CISO's Perspective

Stefan Mindermann, CTO for cybersecurity services at Atos, identified two critical risks: frontier agents operating on hybrid architectures and legacy code being replaced by agentic AI. "As soon as they find a vulnerability in the described process, they can act in any direction," he said, warning that attackers will exploit these weaknesses in libraries and transformation pipelines.

To mitigate these threats, Mindermann recommends a two-layer defense strategy. The first layer focuses on AI governance, where use cases are registered, documented and approved with clear accountability. The second is a security framework that enforces observability, vulnerability detection and real-time penetration ...