Get a free eBook (PDF or ePub) from Manning as well as access to the online liveBook format (and its AI assistant that will answer your questions in any language) when you purchase the print book. Rolling out an AI system without good governance can be really ugly. Data leaks. Biased output. Runaway costs. Poor legal compliance. Privacy exposures. Every company putting AI into production faces these very real risks. This is a one-of-a-kind playbook that translates abstract governance theory into the concrete practices you need for safe, responsible AI. You'll begin with a look at common deployment scenarios and consider how those choices affect control, accountability, and risk. Then, you'll walk through a six-level generative AI governance framework that provides a reusable approach to setting policy, assessing risk, reviewing designs, testing before launch, monitoring live systems, learning from incidents, and other core production concerns. In AI Governance, you'll learn how to: - Match the right safeguards to your deployment model - Apply a framework that transforms governance from paperwork into workflow - Catch problems like prompt injection, bias, and data leakage early - Adapt to AI laws and standards with defensible, scope-aware controls - Operationalize ethics into testing and oversight for trustworthy outcomes The massive training datasets and generalized knowledge that make LLMs like GPT, Gemini, DeepSeek, and Claude so impressive can be a challenge to govern in a production business environment. You want your chatbot to be smart and friendly, but you need to be certain its responses are accurate and relevant. You want to create personalized customer experiences, but you don't want to leak PPI. You want to move fast, but without running afoul of rapidly-changing legal requirements. Establishing a solid plan for Governance, Risk, and Compliance (GRC) in Generative AI projects from the beginning will ensure you reap the benefits of AI without risking the core values of your business. About the book AI Governance: Secure, privacy-preserving, ethical systems presents a structured playbook for safely harnessing the potential of Generative AI, including security and privacy, bias, ethics, cost management, and regulation. Using real-world case studies, you'll explore how hallucinations, bias, and prompt injection manifest and how to mitigate them, including handling new attack vectors like prompt- and data-poisoning. The result is a stable set of guardrails that help you manage what AI can say and do. About the reader For engineering managers, security engineers and architects, privacy officers, and others looking to ship AI to production. About the author Dr. Engin Bozdag leads AI security and privacy design at a global tech company, pioneering LLM security tools, threat modeling, privacy platforms, and GenAI guardrails. Engin is a contributor to ISO 31700 (Privacy by Design) and has trained thousands of engineers in real-world privacy engineering. Dr. Stefano Bennati leads responsible AI and privacy engineering at a global data company He carries out AI Impact Assessments, builds anonymization tools, and implements governance frameworks for ISO 27701 and ISO 42001.
