Week 7

Day 1: Monday, March 2

Topics 6.1-6.2: AI/ML Lifecycle Vulnerabilities and ML Attack Vectors

Topics

• 6.1 AI/ML lifecycle vulnerabilities: data poisoning, backdoors, adversarial examples, model extraction
• 6.2 ML attack vectors across the 4-stage pipeline: data collection, training, deployment, monitoring

Presentation

Presentation 10: AI/ML Lifecycle Vulnerabilities and ML Attack Vectors

Day 2: Wednesday, March 4

Topics 6.3-6.4: LLM-Specific Vulnerabilities and AI/ML Threat Modeling Frameworks

Topics

• 6.3 LLM-specific vulnerabilities: data-control path insecurity, context limits, hallucination, sycophancy, deception
• 6.4 AI/ML threat modeling frameworks: OWASP Top 10 for LLMs, MITRE ATLAS, CSA MAESTRO, AIUC-1

Presentation

Presentation 11: LLM-Specific Vulnerabilities and AI/ML Threat Modeling Frameworks

Reading, Assignments, and Resources