AI Security: Defending LLM Applications (OWASP LLM Top 10, Guardrails, Red-Teaming)

A premium, advanced course on defending large language model applications, updated for 2026. You will learn the modern LLM threat landscape and how to think like a defender, the OWASP Top 10 for LLM Applications in depth (prompt injection, insecure output handling, sensitive information disclosure, supply chain, excessive agency, unbounded consumption and more), layered defenses against direct and indirect prompt injection, input and output guardrails with real tools (Llama Guard, NeMo Guardrails, Lakera, Rebuff), robust system-prompt design, securing RAG pipelines, tool-using agents and Model Context Protocol servers against excessive agency, PII detection and data-leakage prevention, authorized red-teaming and jailbreak defense on your own systems, secrets and supply-chain hygiene, model denial-of-service protection, guardrail evaluation, observability, incident response for AI, and compliance with GDPR and the EU AI Act. Every technique is framed defensively: you protect systems you own or are explicitly authorized to test, and you practice responsible disclosure. Includes a comprehensive final assessment.