SEC 400: AI Security
SEC 400: AI Security equips you to secure modern AI systems. Moving beyond legacy frameworks, this course maps adversarial tactics to MITRE ATLAS and aligns vulnerabilities with the OWASP Top 10 for LLM Applications. You will examine risks across architecture design, model training, and deployment, covering prompt injection, data poisoning, adversarial inputs, and model extraction.
Crucially, you will analyze agentic runtime threats—including Model Context Protocol (MCP) flaws and the "confused deputy" problem. Through hands-on exercises and real-world case studies, you will learn to implement defenses like hardware enclaves, cryptographic verification, and robust prompt protections.
Modules:
Foundations & Frameworks: Dual-use AI, extended CIA triad, and MITRE ATLAS
Architectures: Foundation models, RAG, and OWASP Top 10 baselines
Protocol Vulnerabilities: Analyzing MCP flaws, tool-use security, and privilege escalation
Adversarial Methods
Prompt Exploits
Defenses
Course•By Binu Panicker