01 - Introduction to RAG Security & Threat Model
Understand the architectural mechanics, vector space mathematics, trust boundaries, and threat landscape of Retrieval-Augmented Generation (RAG) systems.
02 - RAG Poisoning & Indirect Injection Mechanics
Master the mechanics of Indirect Prompt Injection, PDF font hiding, zero-width Unicode vectors, semantic clustering data poisoning, and split-chunk attacks in RAG.
03 - Multi-Tenancy & Vector Database Access Control
Implement enterprise multi-tenancy, Row-Level Security (RLS), metadata filtering, namespace isolation, and ABAC in Pinecone, Qdrant, Milvus, and pgvector.
04 - Retrieval Sanitization, Guardrails & Defensive Patterns
Build production-grade RAG defenses: Dual-LLM Quarantine, structural XML tags, NeMo Guardrails, zero-width character stripping, and output exfiltration filters.
05 - RAG Security Evaluations, Red Teaming & CI/CD Security
Automate RAG red-teaming, prompt injection scanning, and multi-tenant isolation testing using Garak, Ragas, PyRIT, Promptfoo, and GitHub Actions.
06 - Hands-on Lab: Exploiting & Securing RAG Applications
Step-by-step vulnerability lab: Build a vulnerable multi-tenant RAG app, execute indirect prompt injection & cross-tenant data leakage, and apply verified defenses.
07 - References & Further Reading
Comprehensive references, academic papers, OWASP standards, MITRE ATLAS framework, vendor security guides, and open-source evaluation tools for RAG security.
Retrieval-Augmented Generation (RAG) Security Guide
Master Retrieval-Augmented Generation (RAG) security: threat modeling, indirect prompt injection, vector database multi-tenancy, data poisoning, and retrieval guardrails.