Certified Ethical Hacker (CEH): Unit 8

Overview

This course is designed to equip you with the knowledge to protect large language models (LLMs) and AI systems from emerging threats. You will explore critical security challenges such as prompt injection, training data poisoning, and model theft. You will gain insights into frameworks like MITRE ATLAS and NIST, and learn to implement best practices for securing AI ecosystems. By the end of this course, you will be proficient in identifying vulnerabilities, applying mitigation strategies, and enhancing the resilience of AI systems.

Syllabus

Securing Generative AI

This module covers securing generative AI. It begins with an introduction to AI threats and large language model (LLM) security. You will learn about OS Top 10 for LLM applications and the MITRE ATLAS framework. You will learn about the Coalition for Secure AI and the best practices being developed by organizations like NIST and others. You will learn about prompt injection, insecure output handling, training data poisoning, model denial of service, and supply chain security. You'll also learn about other threats, like sensitive information disclosure, insecure plugin design, and excessive agency. You will learn concepts that will help you understand overreliance in AI, model theft attacks, and understanding red teaming of AI models. The module will also cover retrieval-augmented generation (RAG) and its different permutations, as well as explore tools like LangChain, LlamaIndex, LangGraph, and other orchestration libraries used with AI. You will learn how to secure embedding models, secure vector databases, and develop strategies for monitoring and incident response.