Defending AI Systems: LLM and Agent Security

As AI systems and LLM agents become deeply integrated into modern software, securing them against unique vulnerabilities is more critical than ever. This course introduces you to the essential concepts of AI security, helping you understand how attackers exploit these systems and how to defend them. You will transition from having a basic understanding of AI to being able to identify, analyze, and mitigate security risks across the entire lifecycle of an AI application. What you'll learn: Understand the foundational security vulnerabilities inherent in Large Language Models and neural networks; Analyze common threat vectors including prompt injection, jailbreaking, and data poisoning; Explore industry-standard security frameworks such as the OWASP Top 10 for LLM Applications and MITRE ATLAS; Evaluate risks unique to autonomous AI agents, including tool poisoning and supply chain vulnerabilities; Apply defensive engineering techniques to protect models, memory, data, and system architectures; Practice identifying vulnerabilities through structured written scenarios and conceptual exercises. The course starts with basic terminology and foundational AI concepts before guiding you through real-world attack vectors and modern defensive strategies. You will read detailed breakdowns of security incidents and analyze defensive architectures step-by-step. This course is designed for software developers, security beginners, and IT professionals looking to enter the field of AI security, with no prior background in security required. Start reading today to build a strong foundation in protecting modern AI systems.