Compendium Defense

Defense

23 lessons covering defense concepts and techniques.

intermediate 10 minutes

AI Coding Tool Security — Defending Development Assistants

AI coding tools require specialized security controls beyond general LLM defenses due to their unique combination of cod...

intermediate 8 minutes

ATLAS-Informed Defense Planning: From Attack Mapping to Mitigation Strategy

Learn how to use MITRE ATLAS proactively to anticipate attack paths, prioritize mitigations, and close security gaps bef...

intermediate 10 minutes

Automated Red Teaming as a Defensive Flywheel: Building Sustainable Agent Security

Learn why one-time hardening is insufficient for agent security and how automated red teaming creates a sustainable defe...

intermediate 10 minutes

Automated Red Teaming as a Defensive Practice

Learn how automated red teaming creates a continuous defensive flywheel that systematically discovers vulnerabilities an...

advanced 9 minutes

Confabulation as Defense: When Hallucination Protects Secrets

Learn how defensive confabulation can satisfy extraction pressure with plausible fake content while protecting real secr...

intermediate 9 minutes

Confirmation Gates and Constrained Actions: Limiting Agent Risk Through System Controls

Learn why agent safety depends less on blocking every prompt injection and more on constraining what actions an agent ca...

intermediate 10 minutes

Data and Model Poisoning — Protecting AI Integrity from Training to Deployment

Data and model poisoning attacks manipulate training data or model weights to introduce vulnerabilities, backdoors, or b...

intermediate 8 minutes

Defense Strategy Core Principles

Learn the core strategic principles behind robust AI defense, from defense in depth and intent recognition to secure fai...

intermediate 9 minutes

FIDS: Foreign Instruction Detection Through Separation — Training-Time Defenses for Document Pipelines

Learn why training-time defenses using LoRA adaptation can outperform inference-time mitigations, and how Foreign Instru...

intermediate 8 minutes

Improper Output Handling: Validating LLM Outputs Before They Reach Your Systems

Learn why treating LLM outputs as untrusted input and applying proper validation, encoding, and sanitization prevents XS...

intermediate 9 minutes

Intent Security and Behavioral Monitoring for AI Agents

Learn how intent security and behavioral monitoring provide a proactive defense layer for AI agents by observing what th...

intermediate 10 minutes

Measuring AI Security Risk: Metrics and Methods

Learn how to quantify AI security risks through the NIST AI RMF Measure function — using concrete metrics, tracking indi...

intermediate 10 minutes

Multi-Language Safety Evaluation: Defending Against Cross-Lingual Jailbreaks

Learn why language-specific safety filters create exploitable blind spots and how to design robust multi-language evalua...

advanced 9 minutes

Defending Against Multi-Vector Jailbreak Attacks

Learn how robust defenders withstand compound jailbreak attempts that combine formatting, roleplay, authority claims, an...

intermediate 9 minutes

When PDF Prompt Injection Hits Production — A Remediation Playbook

Learn the structured response process engineering teams need when PDF prompt injection is reported in production, coveri...

intermediate 7 minutes

Prompt Drift as a Defensive Signal: Monitoring System Prompt Changes

Learn how to use system prompt version changes as a defensive observability signal to detect unauthorized modifications,...

intermediate 10 minutes

Securing RAG Pipelines: Defense Against Knowledge Base Attacks

Learn how to defend Retrieval-Augmented Generation (RAG) systems against knowledge base poisoning, retrieval manipulatio...

intermediate 10 minutes

Resource Exhaustion Detection: Preventing Computational DoS in LLM Applications

Learn how encoder models can predict LLM output length and computational cost before generation, enabling proactive defe...

intermediate 8 minutes

SAIF Automated Defenses — Scaling Security to Match Threat Velocity

Learn why manual security testing cannot keep pace with AI threats and how automated defenses—adversarial testing, conti...

intermediate 8 minutes

Harmonizing Platform Controls — Consistent Security Across AI Environments

Learn how to maintain consistent security controls across diverse AI platforms and deployment environments to prevent at...

intermediate 8 minutes

Sensitive Information Disclosure: Preventing LLMs from Leaking Secrets

Learn why LLMs can inadvertently disclose sensitive information and how layered defenses from training data through outp...

intermediate 10 minutes

Tool Calling and Agent Security Best Practices

Learn defensive patterns for securing AI agent tool-use capabilities against prompt injection attacks and unauthorized a...

intermediate 10 minutes

Securing RAG Systems: Defending Against Vector and Embedding Weaknesses

Learn to defend Retrieval-Augmented Generation (RAG) systems against vector and embedding weaknesses that can lead to da...