AI Security Fundamentals
A practical introduction to AI security, vulnerabilities, threat modeling, and securing modern AI systems.
AI Security Fundamentals is a hands-on course for developers, security professionals, and technical leaders who need to understand how modern AI systems — especially LLM-based applications and agentic AI — can be attacked, and how to defend them.
You'll learn how to think like both an attacker and a defender: modeling threats against real AI architectures, running prompt injection and red-teaming exercises against a sample application, and applying practical, low-friction mitigations that hold up in production.
What you'll learn
- Understand the unique security properties of LLMs and agentic AI systems
- Model threats against a real-world AI application architecture
- Identify and reproduce prompt injection vulnerabilities
- Apply practical mitigations: least privilege, approval gates, and monitoring
- Run a lightweight AI red-teaming exercise
Curriculum
1. Why AI systems are different
How LLMs blend instructions and data, and what that means for security.
2. Threat modeling for AI systems
Mapping attack surfaces across prompts, tools, RAG pipelines and agents.
3. Prompt injection deep dive
Hands-on exploration of direct and indirect prompt injection.
4. Data leakage & excessive agency
How AI systems can be tricked into leaking data or taking unwanted actions.
5. Red teaming & secure deployment
Practical techniques to test and harden AI applications before shipping.