AI Security Engineer
The fastest-growing specialty in security. Learn to identify and mitigate risks unique to AI/ML systems: from prompt injection to model poisoning and AI governance.
Who this path is for
This path is for security engineers, application security staff and cloud engineers whose companies are rolling out chatbots, copilots or AI agents. You should already understand web application security and basic cloud architecture. The target role is someone who can threat model an AI feature, test it, and help set the rules for using it safely.
What you'll be able to do
- Explain the main LLM risks, including prompt injection, insecure output handling and sensitive data leaking through responses.
- Plan and run a basic AI red team exercise against a chatbot or agent, and report what you found.
- Identify risks to models and training data such as poisoning, model theft and supply chain issues in model files.
- Map an AI system to governance frameworks like the NIST AI RMF so risk owners can make decisions.
- Recommend practical guardrails: input and output filtering, least privilege for agent tools and logging of AI activity.
What you'll cover
LLM Security Fundamentals
Understand the attack surface of Large Language Models: prompt injection, jailbreaks, data leakage, and insecure output handling.
AI Red Teaming
Learn how to systematically test AI systems for safety and security failures before adversaries do.
Model & Training Security
Protect the ML pipeline from data poisoning, model stealing, and backdoor attacks.
AI Governance & Compliance
Navigate the EU AI Act, NIST AI RMF, and build internal governance for responsible AI deployment.
Knowledge Check
Related certifications
Protego AI Security course
A free course that goes deeper into AI threats, governance frameworks, agentic AI risks and AI red teaming.
Free AI Security course →Microsoft Cloud and AI Security Engineer Associate
The exam that replaced AZ-500, which retired on August 31, 2026. It includes securing AI workloads on Azure alongside classic cloud security.
Free SC-500 course →Microsoft Azure AI Fundamentals
A light introduction to Azure AI services. Useful if you need to understand the platforms before you can secure them.
Frequently asked questions
How long does the AI Security Engineer path take?
Plan for about 10 hours across the four topic areas and the quiz. Practicing prompt injection against a test chatbot adds valuable hands-on time on top of that.
Do I need a machine learning background?
No. You need solid security fundamentals and a working understanding of how applications call an AI model. The path explains the AI specific parts as it goes.
What job titles does this path lead to?
Relevant titles include AI security engineer, AI red teamer, application security engineer with an AI focus and security architect for AI platforms.
Path Summary
- Level
- Intermediate
- Estimated time
- 10 hours
- Topics
- 4
- Resources
- 12
- Quiz questions
- 10
- Passing score
- 70% (7/10)