Chapter 2

Adversarial Attacks

You will be able to classify attacks by lifecycle stage and attacker access, explain how FGSM and physical attacks work, trace a model-extraction pipeline, and distinguish poisoning, prompt injection, and sycophancy.

• 1. A Map of Adversarial Attacks→
• 2. White-Box and Black-Box Attacks→
• 3. Evasion and the Fast Gradient Sign Method→
• 4. One-Pixel Attacks→
• 5. Adversarial Patch Attacks→
• 6. Extraction and Inference Attacks→
• 7. Poisoning and Backdoors→
• 8. Prompt Injection and Sycophancy→