Lilly Tech Systems
AI Threat Modeling
Learn to systematically identify, assess, and prioritize security threats targeting AI and machine learning systems. Master frameworks like STRIDE adapted for AI, understand the OWASP ML Top 10, and build comprehensive threat models that protect your AI deployments from adversarial attacks, data poisoning, and model exploitation.
What You'll Learn
By the end of this course, you'll be able to build comprehensive threat models for any AI system and implement effective risk mitigation strategies.
AI-Specific Threats
Understand threats unique to AI systems including adversarial attacks, model theft, data poisoning, and prompt injection.
Risk Assessment
Learn structured risk assessment frameworks to evaluate threat likelihood, impact, and prioritize mitigations effectively.
STRIDE for AI
Apply the STRIDE methodology adapted for machine learning systems, covering spoofing, tampering, repudiation, and more.
Mitigation Strategies
Build defense-in-depth strategies with input validation, model hardening, monitoring, and incident response planning.
Course Lessons
Follow the lessons in order to build a complete understanding of AI threat modeling.
1. Introduction
What is AI threat modeling? Learn why traditional threat modeling falls short for AI systems and discover frameworks designed for ML security.
2. Threat Landscape
Explore the AI threat landscape including OWASP ML Top 10, MITRE ATLAS, and real-world attack case studies on production AI systems.
3. STRIDE for AI
Apply the STRIDE threat modeling framework to AI systems. Learn how each category maps to AI-specific vulnerabilities and attack vectors.
4. Attack Surfaces
Map the full attack surface of AI systems: training pipelines, model serving, APIs, data stores, and supply chain dependencies.
5. Mitigation
Implement defense-in-depth strategies including input validation, adversarial training, model monitoring, and security controls.
6. Best Practices
Enterprise-grade threat modeling processes, continuous assessment, compliance frameworks, and building a security-first AI culture.
Prerequisites
What you need before starting this course.
- Basic understanding of machine learning concepts
- Familiarity with general cybersecurity principles
- Understanding of software development lifecycle
- No coding required — this course focuses on frameworks and methodology