Security for AI in AWS by Miguel Brito

Generative AI has introduced an entirely new security landscape. Traditional cloud security remains essential, but protecting AI systems now requires addressing threats such as prompt injection, excessive agent autonomy, data leakage, model misuse, insecure RAG pipelines, and AI governance. Security for AI in AWS provides a practical, architecture-focused approach to securing Generative AI, agentic AI, and machine learning workloads running on AWS. Inside this book you'll learn how to: • Design least-privilege IAM for AI workloads • Secure Amazon Bedrock and RAG architectures • Defend against prompt injection and agent-related threats • Protect data throughout the AI lifecycle • Implement Bedrock Guardrails effectively • Build secure multi-account and multi-tenant AI platforms • Monitor, audit, and govern AI systems in production Packed with architecture diagrams, implementation patterns, security checklists, and real-world examples, this guide helps you build AI systems that are secure, scalable, and production-ready. Whether you're designing your first Generative AI application or securing enterprise AI platforms, this book provides the practical guidance needed to build trustworthy AI solutions on AWS.