Course Prerequisites:
Basic Cybersecurity Knowledge: Understanding of fundamental security concepts and protocols.
Networking Fundamentals: Familiarity with networking concepts and protocols.
Risk Management Experience: Experience with risk assessment and mitigation strategies.
System Integration Skills: Understanding of integrating complex IT systems.
Familiarity with Compliance Standards: Knowledge of regulatory frameworks and industry standards.
Modules:
Module 1: AI Security Foundations – Responsible Development & Secure Design
1.1 Overview of AI Security Challenges
1.2 Secure AI Design Principles
1.3 Best Practices for Secure AI
1.4 RSAIF MOSAIC Domains and Mitigation Alignment
1.5 Core Privacy Attacks on ML Models
1.6 LLM-Specific Attack Vectors
1.7 AI Attack Surface Diagram
1.8 Safety & Alignment Failure Modes
1.9 Hands-On (No Code)
Module 2: AI Threat Models
2.1 Introduction to Threat Modeling
2.2 Overview of Creating an AI Threat Model
2.3 Tools for Threat Modeling
2.4 Case Study: AI in Autonomous Vehicles
2.5 Training Time vs. Inference Time Attacks
2.6 Adversarial Robustness: Metrics and Evaluation Frameworks
2.7 Practitioner-Ready Threat Modeling Templates for LLM Systems
2.8 Real Adversary Behavior Sequences in the Wild
2.9 Hands-on
Module 3: Secure AI SDLC (Software Development Lifecycle)
3.1 Overview of Software Development Life Cycle (SDLC)
3.2 AI-Specific Security Measures
3.3 Overview: Continuous Monitoring & Feedback Loops
3.4 Expanding CI/CD, Model Registry, Reproducibility, and Artifact-Signing
3.5 Use Case: AI Fraud Detection System
3.6 Overview: Securing the AI/ML Model Supply Chain
3.7 Data Provenance and Lineage Controls
3.8 Secure RAG Pipeline Requirements
3.9 GPU Runtime Security, Memory Isolation, and Co-Tenant Inference Protection
3.10 Reproducibility Requirements for Regulated Industries
3.11 GRC-Focused Interpretation of the Secure AI SDLC
3.12 RSAIF Mapping Across the Secure AI SDLC
3.13 Hands-on
Module 4: Enforcement & Model Integrity
4.1 Securing AI Systems Post-Deployment
4.2 Model Integrity and Auditing
4.3 Cryptographic Integrity Protections (Hash Validation & Signature Rotation)
4.4 Side-Channel Attack Scenarios on Model Checkpoints, Quantized Models, and GPU
4.5 Guardrail Testing Patterns for Automated Prompt Sanitization
4.6 Separation of Duties & Dual Control for High-Risk AI Models
4.7 Evaluation Guidance for Model Behavior Consistency
4.8 RSAIF Mapping, GRC Interpretation, and Evidence Requirements
4.9 Introducing Dual Lab Paths and a Tools Capability Matrix
4.10 Hands-On: Implementing RBAC for Secure AI APIs (Dual Lab Path)
4.11 Knowledge Check
Module 5: Audit Readiness & Red-Teaming
5.1 Overview on Preparing AI Systems for Audits
5.2 Overview on Red-Teaming for AI Systems
5.3 Regulator-Required Documentation for EU AI Act, ISO 42001 Conformity, and NIST
5.4 Post-Market Monitoring & Incident Reporting
5.5 Deeper Red Team Scenarios: Latent Space Attacks, Jailbreak Escalation Chains,
5.6 AI Incident Response Playbook
5.7 Red Team Scoring System and Pass/Fail Criteria
5.8 Audit Artifact Lifecycle, Required Documentation, and Compliance Mapping
5.9 Red-Team Testing for Modern AI Systems
5.10 Complete Red-Team Workflow and Evidence Templates for AI Systems
5.11 RSAIF Alignment and Testing Frameworks
5.12 No-Code and Low Code Lab Exercise
5.13 Knowledge Check
Module 6: Toolkits & Automation
6.1 Security Components – Tooling & Automation (RSAIF-Aligned)
6.2 Automating AI Security and Compliance
6.3 Safety Controls – Hallucination Monitoring & Scoring
6.4 Architecture of Automated Compliance Pipeline
6.5 Automated Rollback Workflows, Drift Alerts, and Scheduled Red Teaming
6.6 Cross-Model Validation for Multi-Model AI Systems
6.7 GPU Runtime Observability and Isolation Requirements
6.8 Introduction: AI Security Automation Stack
6.9 Expanding AI Security Tool Categories
6.10 Tool Selection Criteria and Capability Matrix
6.11 Real Automation Workflow & Evidence Generation
6.12 Hands-On Lab