A practical, current course on securing generative-AI systems, built on the OWASP Top 10 for LLM Applications and agentic-AI security. Designed for security engineers, developers, and architects, it covers how LLMs create a new attack surface; prompt injection and jailbreaks (direct and indirect) and their layered defenses; sensitive-information disclosure and system-prompt leakage; data/model poisoning and AI supply-chain security; RAG and vector/embedding weaknesses; improper output handling and misinformation; agentic risks (excessive agency, tool misuse, memory poisoning, unbounded consumption); and governance, red-teaming, and secure MLSecOps using NIST AI RMF, ISO/IEC 42001, and MITRE ATLAS. Every domain has plain-language lessons, concept tables, exam-trap warnings, hands-on labs, a knowledge check, and a randomized mastery check, ending with a full-length mock and a spaced-repetition flashcard deck. Vendor-neutral. Free.