Skill guide

AI security

AI security protects model-enabled applications from risks including prompt injection, unsafe tool use, data leakage, poisoned retrieval, and excessive permissions.

Why this capability matters

AI systems can convert untrusted text into actions, so security must cover data, instructions, tools, identity, monitoring, and human authority.

What competent practice includes

Threat modelling
Prompt-injection testing
Least-privilege tool design
Input and output controls
Security review and monitoring
Portfolio evidence

A threat model, red-team test set, mitigations, residual-risk assessment, and security handover for a sample AI application.

See public project briefs

Professional applications

These are fields of application, not guaranteed job or income outcomes.

Agent security reviewsRAG threat assessmentsGovernance controlsRed-team exercisesSecure deployment guidance

Courses that develop this skill

AI Security
AI Security
Defending LLM applications and agents against prompt injection, jailbreaks, and the OWASP Top 10 for LLMs.
AI Security
AI LLM Testing
Quality engineering for AI systems — testing non-deterministic LLM and RAG outputs the way QA tests deterministic code.

Connect the skill to a market path

Learn how this capability fits inside a complete problem, proof, service, and delivery journey.

Read the connected guide