WSHA AI Security Series: Machine-Speed Threats, Human-Speed Defenses

Artificial intelligence has shifted from being a topic on the security agenda to the driving force reshaping every aspect of it.

We begin by exploring how attackers leverage AI today: from deepfake-enabled email compromises and fraudulent remote employees to autonomous agents conducting full-scale attack campaigns at machine speed. We then address, with unvarnished honesty, what AI-powered defenses can and cannot do. Then, we turn to real-world penetration testing of AI-augmented networks, where vulnerabilities are often surprisingly mundane: unhardened systems, default credentials, and unpatched services.

Later sessions tackle the risks organizations create for themselves—Shadow AI and ungoverned data flows, AI-generated code left unreviewed, agentic tools and browser agents quietly granted extensive access to corporate systems, and vendor-added AI features introduced without transparency.

The series concludes with a look at a critical, often overlooked domain: treating AI systems as targets, securing and testing your AI infrastructure, and establishing governance, accountability, and regulatory frameworks. We finish with a practical 90-day action plan and a maturity model ready for board presentation.