References
Useful AI security resources — third-party links are provided for further learning and attribution.
Original content
This resource is original content by GenAISecurityLab. Third-party resources are linked for further learning and attribution. We do not copy third-party payload lists or protected content.
Resources
OWASP OWASP Top 10 for LLM Applications
The OWASP project's GitHub repository — source docs and version history behind the LLM Top 10, not the published list itself.
OWASP OWASP LLM Top 10
The published, citable OWASP Top 10 for LLM Applications list, hosted on the OWASP GenAI site.
OWASP OWASP AI Testing Guide
Open OWASP project: a structured methodology and test catalogue (AITG test IDs) for AI systems, spanning model, data, and application-level testing.
OWASP OWASP LLM Prompt Injection Prevention Cheat Sheet
Defensive guidance for prompt-injection prevention.
OWASP OWASP AI Agent Security Cheat Sheet
Defensive guidance for AI agent security.
NIST NIST Adversarial Machine Learning Taxonomy
NIST taxonomy for adversarial machine learning concepts.
OWASP OWASP AI Vulnerability Scoring System (AIVSS)
A CVSS-style scoring framework for rating the severity of AI-specific vulnerabilities — model flaws, data weaknesses, and deployment exposure.
OWASP OWASP GenAI Red Teaming Guide
A structured, risk-based methodology for red teaming generative-AI systems, from model evaluation to runtime behaviour and system-integration pitfalls.
CSA CSA Agentic AI Red Teaming Guide
Cloud Security Alliance guidance for red teaming autonomous, goal-driven AI agents and their agentic attack surface.
Tooling garak
NVIDIA's open-source LLM vulnerability scanner — automated probes for prompt injection, jailbreak, toxicity, and data leakage.
Tooling Promptfoo
Open-source LLM evaluation and red-team harness; runs adversarial and regression test suites in CI/CD.
Tooling DeepTeam
Open-source framework for red teaming LLMs and AI agents with reusable attack modules and vulnerability suites.
Tooling PyRIT (Python Risk Identification Toolkit)
Microsoft's Python red-teaming toolkit for generative AI. Note: the repository was archived in 2026 (read-only) but remains widely referenced.
Keep learning in a controlled lab.
Practice prompt injection, RAG leakage, and AI agent tool abuse in authorized environments before applying the methodology to real reviews.