AI Security & Governance Masterclass: Build, Attack & Defend

About this course

This course contains the use of artificial intelligence.

Build the practical skills required to design, attack, secure, monitor, and govern modern AI applications in this comprehensive AI Security and Governance Masterclass.

As organizations rapidly adopt generative AI, large language models, RAG applications, and autonomous AI agents, security and governance have become essential. AI systems introduce risks that traditional application security alone cannot address, including prompt injection, jailbreak attacks, document poisoning, memory manipulation, unauthorized tool execution, sensitive data exposure, hallucinations, and uncontrolled agent actions.

This course takes a hands-on, project-based approach. You will not only study AI security concepts—you will build vulnerable AI systems, attack them, understand their weaknesses, and implement practical defenses.

You will begin by setting up Python, Visual Studio Code, Git, and Ollama, then build your first AI chat assistant. You will explore the AI threat landscape, understand the OWASP Top 10 for LLM Applications, and learn how security differs across traditional software and AI-powered systems.

You will perform real-world direct prompt injection and AI jailbreak attacks, then create prompt validation, risk scoring, filtering, and guardrail mechanisms. You will build a complete Retrieval-Augmented Generation application, attack it using malicious documents, poison retrieved context, and secure it through trusted-source validation, context isolation, content sanitization, and enforcement controls.

The course also covers AI tool-calling security, including parameter injection

Prices, discounts and availability are set by Udemy. We may earn a commission when you purchase through links on this site.