AI/ML Security Course

Attack and defend AI and ML systems.

19 modules and 226 topics, taught step by step with simple lessons, diagrams, labs and bug challenges.

Introduction to AI/ML Security

  • Introduction to Artificial Intelligence
  • AI vs Machine Learning vs Deep Learning
  • Generative AI and Large Language Models
  • What Is AI Security?
  • AI Security vs Traditional Cybersecurity
  • AI Threat Landscape
  • AI Attack Surface and Trust Boundaries
  • AI Security Principles
  • Real-World AI Security Incidents
  • AI Security Roles and Career Paths

Machine Learning Fundamentals for Security

  • Supervised Learning
  • Unsupervised Learning
  • Reinforcement Learning
  • Classification and Regression
  • Clustering and Anomaly Detection
  • Neural Networks Fundamentals
  • Deep Learning Architectures
  • Natural Language Processing
  • Computer Vision Fundamentals
  • Model Training, Validation and Testing
  • Overfitting and Underfitting
  • Model Evaluation Metrics

AI/ML Architecture and Lifecycle

  • AI System Architecture
  • Machine Learning Lifecycle
  • Data Collection and Processing
  • Feature Engineering
  • Model Training Pipelines
  • Model Validation and Evaluation
  • Model Serialization and Storage
  • Model Deployment Architecture
  • Model Serving and Inference
  • Model Versioning
  • ML Pipeline Security
  • AI System Trust Boundaries

AI Data Security, Privacy and Governance

  • Data Security Fundamentals
  • Training Data Classification
  • Personally Identifiable Information (PII)
  • Sensitive Data Detection
  • Data Anonymization and Pseudonymization
  • Data Encryption at Rest and in Transit
  • Data Leakage Risks
  • Dataset Access Control
  • Data Retention and Deletion
  • Privacy-Preserving Machine Learning
  • Differential Privacy Fundamentals
  • Data Governance and Compliance

Adversarial Machine Learning

  • Introduction to Adversarial Machine Learning
  • Adversarial Threat Models
  • Adversarial Examples
  • Evasion Attacks
  • White-Box and Black-Box Attacks
  • Targeted and Untargeted Attacks
  • Adversarial Perturbations
  • Image Classification Attack Scenarios
  • Text Classification Attack Scenarios
  • Adversarial Robustness Evaluation
  • Adversarial Training
  • Defensive Strategies

ML Model Attacks and Threats

  • Model Extraction Attacks
  • Model Stealing Techniques
  • Model Inversion Attacks
  • Membership Inference Attacks
  • Model Fingerprinting
  • Model Replication Risks
  • Training Data Reconstruction
  • Prediction API Abuse
  • Rate Limiting and Query Abuse
  • Model Intellectual Property Protection
  • Model Access Control
  • Model Attack Detection

Training Data Poisoning and Supply Chain Security

  • Training Data Integrity
  • Data Poisoning Fundamentals
  • Targeted and Untargeted Poisoning
  • Label Flipping Attacks
  • Backdoor Attacks in ML Models
  • Trigger-Based Model Manipulation
  • Dataset Manipulation Risks
  • Third-Party Dataset Security
  • Pretrained Model Supply Chain Risks
  • Model Repository Security
  • Dependency and Package Integrity
  • Dataset and Model Provenance

ML Model Security and Robustness

  • Secure Model Design
  • Model Robustness Fundamentals
  • Adversarial Robustness Testing
  • Input Validation and Sanitization
  • Out-of-Distribution Detection
  • Model Explainability and Interpretability
  • Model Bias and Fairness Risks
  • Model Privacy Protection
  • Secure Model Serialization
  • Model Integrity Verification
  • Model Access Restrictions
  • Secure Model Evaluation

Deep Learning Security

  • Deep Learning Security Fundamentals
  • Neural Network Attack Surfaces
  • CNN Security
  • RNN and LSTM Security
  • Transformer Architecture Security
  • Vision Model Security
  • Speech and Audio Model Security
  • Multimodal AI Security
  • Adversarial Inputs in Deep Learning
  • Model Weight Security
  • Gradient Leakage
  • Deep Learning Defense Techniques

Large Language Model Fundamentals

  • Introduction to LLMs
  • Transformer Architecture
  • Tokens and Tokenization
  • Embeddings
  • Attention Mechanism
  • Pretraining and Fine-Tuning
  • Instruction Tuning
  • RLHF Fundamentals
  • Open-Source vs Closed-Source Models
  • LLM Inference Architecture
  • Context Windows and Memory
  • LLM APIs and Model Serving

LLM Security Threat Landscape

  • Introduction to LLM Application Security
  • OWASP Top 10 for LLM Applications
  • LLM Threat Modeling
  • Prompt Injection
  • Sensitive Information Disclosure
  • Supply Chain Vulnerabilities
  • Data and Model Poisoning
  • Improper Output Handling
  • Excessive Agency
  • System Prompt Leakage
  • Unbounded Consumption
  • LLM Security Testing Methodology

Prompt Injection and Jailbreak Security

  • Prompt Injection Fundamentals
  • Direct Prompt Injection
  • Indirect Prompt Injection
  • System Prompt Manipulation
  • Jailbreak Concepts
  • Instruction Hierarchy
  • Prompt Leakage Risks
  • Malicious Document and Web Content
  • Multimodal Prompt Injection
  • Input and Output Guardrails
  • Prompt Injection Detection
  • Defensive Prompt Engineering

Retrieval-Augmented Generation (RAG) Security

  • RAG Architecture Fundamentals
  • Document Ingestion Security
  • Data Chunking and Embeddings
  • Vector Database Security
  • Retrieval Access Control
  • Cross-Tenant Data Leakage
  • RAG Data Poisoning
  • Malicious Document Injection
  • Retrieval Manipulation
  • Sensitive Data Exposure
  • Secure RAG Design
  • RAG Security Testing and Validation

AI Agent and Tool-Use Security

  • AI Agent Architecture
  • Autonomous AI Agents
  • Agent Planning and Execution
  • Tool Calling and Function Calling
  • Agent Memory Security
  • Agent Permission Models
  • Excessive Agency Risks
  • Tool Injection Attacks
  • Indirect Prompt Injection in Agents
  • Multi-Agent Security
  • Human-in-the-Loop Controls
  • Secure Agent Architecture

AI API and Application Security

  • AI API Architecture
  • LLM API Authentication
  • API Key and Secret Management
  • API Authorization
  • Rate Limiting and Quotas
  • Input Validation
  • Output Validation
  • Insecure API Exposure
  • AI API Abuse and Cost Attacks
  • Multi-Tenant AI Security
  • API Logging and Monitoring
  • Secure AI Application Development

AI Infrastructure and Deployment Security

  • AI Infrastructure Architecture
  • GPU and Compute Environment Security
  • Model Hosting Security
  • Inference Server Security
  • Containerized ML Environments
  • ML Pipeline Access Control
  • Secrets and Credential Management
  • Model Artifact Storage Security
  • AI Environment Isolation
  • Secure Deployment Practices
  • AI Infrastructure Monitoring
  • AI Deployment Hardening

AI Red Teaming and Security Evaluation

  • Introduction to AI Red Teaming
  • AI Threat Modeling
  • AI Attack Surface Mapping
  • LLM Security Assessment Methodology
  • Prompt Injection Evaluation
  • Jailbreak Resistance Testing
  • RAG Security Assessment
  • AI Agent Security Testing
  • Model Robustness Testing
  • Data Leakage Assessment
  • Automated AI Security Evaluation
  • Responsible Reporting and Remediation

AI Security Monitoring and Incident Response

  • AI Security Monitoring Fundamentals
  • Model Inference Logging
  • Suspicious Prompt Detection
  • Anomalous API Usage Detection
  • Model Abuse Detection
  • Data Leakage Detection
  • Model Drift and Integrity Monitoring
  • AI Security Incident Classification
  • Incident Response for AI Systems
  • Model Rollback and Recovery
  • Forensic Investigation of AI Incidents
  • AI Security Audit Trails

Secure AI Development and Governance

  • Secure AI Software Development Lifecycle
  • AI Security Requirements
  • AI Risk Assessment
  • AI Threat Modeling Frameworks
  • Secure Dataset Management
  • Model Governance
  • AI Supply Chain Governance
  • Responsible AI Principles
  • AI Compliance Fundamentals
  • NIST AI Risk Management Framework
  • AI System Documentation
  • AI Security Policies and Best Practices