Chapter 6 - Engineering Trust - AI Security, Governance & Guardrails

Chapter 6 focuses on establishing trust as a foundational requirement for enterprise AI by addressing the security, governance, and control mechanisms needed to manage the unique risks of generative AI systems. It examines threats such as prompt injection, data leakage, excessive agency, tool abuse, and AI supply chain risks, while establishing controls for data protection, identity and access, input and output guardrails, policy enforcement, human approval, and auditability. The chapter brings these practices together in the Enterprise AI Governance & Guardrail Matrix, providing a structured foundation for deploying AI securely and responsibly in production.