Dursey P. Red Teaming AI. Attacking and Defending Intelligent Systems 2025

Textbook in PDF format

Inside you will master
Adversarial Tactics - data poisoning, inference‑time evasion, model extraction, LLM prompt injection.
Battle‑hardened Defenses - robust training, MLSecOps pipeline hardening, real‑time detection.
LLM & Agent Security - jailbreak techniques and mitigations for ChatGPT‑style models.
Human‑Factor Threats - deepfakes, AI‑powered social engineering, deception counter‑measures.
STRATEGEMS (TM) Framework - a proprietary, hypergame‑inspired methodology to red‑team AI at scale.
Why trust this guide?
Author Philip A. Dursey is a three‑time AI founder and ex‑CISO who has secured billion‑dollar infrastructures and leads HYPERGAME’s frontier‑security practice.
Who should read
Security engineers * Red teamers * ML/AI researchers * CISOs & CTOs * Product and policy leaders