Page 130 - Cyber Defense eMagazine September 2025

How Red Teams are Reinventing Cybersecurity for the Age of

AI

As generative AI systems permeate critical infrastructure, red teams must adapt their
tactics to confront unpredictable behaviors, data leakage, and multi-modal
vulnerabilities.

By Dr. Darren Pulsipher, Chief Solution Architect, Intel

Red teaming has long served as a cornerstone of cybersecurity, probing networks and platforms for flaws
before attackers can exploit them. Now, these cyber professionals are turning those same tactics to
artificial intelligence systems to help ensure trust and security.

However, testing AI requires different evaluation tactics. Conventional IT systems behave predictably.
Not generative AI. Its non-deterministic nature can produce different results from the same prompt. It also
lacks fixed rules, making it harder for red teams to test and secure consistently.

Ironically, the very unpredictability that complicates red teaming also makes it indispensable, especially
as AI plays a growing role in essential societal functions and public safety. Let’s take a closer look at the
main risks tied to GenAI and the emerging strategies red teams can use to strengthen its security and
reliability.

Cyber Defense eMagazine – September 2025 Edition 130
Copyright © 2025, Cyber Defense Magazine. All rights reserved worldwide.