AI Security Basics

AI security covers the security of machine learning models and AI applications, as well as using AI to solve security tasks.

In progress

This category is being written. Planned topics:

• Machine learning basics and how large language models work
• Prompt injection and jailbreaks
• Adversarial examples and model robustness
• Model extraction and membership inference
• Training data poisoning and backdoors
• RAG, agents and tool-calling security
• Output filtering and sandbox escapes
• AI-assisted reverse engineering, forensics and vulnerability research

Study tips

1. Learn how models are called, how context works and how tools are invoked.
2. Start with prompt injection and jailbreak challenges.
3. Pay attention to AI-specific attack surfaces such as knowledge bases, plugins and agent permissions.