AS-201b

Threat Modeling for AI Systems

AI-specific threat models, expanded attack surface analysis, prompt injection defense in depth, data exfiltration prevention, and the structured methodology that turns security from reactive firefighting into proactive engineering.

10 Lessons · ~0.5 Hours · 3 Modules

Instructor: DRILL — Academy Director

Module 1: AI Attack Surfaces

Mapping the complete threat landscape for AI systems — where traditional security ends, where AI-specific risks begin, and the structured methodology for identifying both.

Module 2: Prompt Injection Defense

Defense in depth against the most exploited AI vulnerability — input hardening, output validation, architectural isolation, and the layered approach that raises the bar from trivial to impractical.

Module 3: Data Exfiltration Prevention

Preventing AI systems from leaking sensitive data — context window hygiene, output guardrails, logging for forensics, and the organizational policies that make prevention systematic.