Building Safer Technology: A Field Guide to Failing Well

Failure is inevitable. Failing well is a skill.

The era of "move fast and break things" is over. In a world where software and AI are woven into everything from power grids to the devices in our homes, failure is no longer a minor inconvenience―it's a critical business risk that can cost lives, paralyze economies, and erode trust. Building Safer Technology delivers a practical field guide to safety engineering, a practice dedicated to designing systems that are not only innovative but also fundamentally resilient.

Go beyond theory and learn to implement safety practices directly into your workflows and organizational structure. This book provides concrete tools, mental models, and communication strategies for individuals and team leaders to build products that can withstand real-world chaos. You'll learn how to prevent failures and also how to manage them gracefully when they inevitably occur, preserving trust and turning crises into powerful learning opportunities.

You will learn how to:

• Learn to apply the Safety Engineering Methodology, a four-step process to understand, analyze, and mitigate risk in any sociotechnical system.
• Identify and analyze potential failure modes in any product, from AI models to infrastructure, using real-world case studies like the Boeing 737 MAX and Zillow’s home-pricing algorithm.
• Develop robust mitigation plans and resiliency strategies that transform potential disasters into manageable, predictable events.
• Establish effective incident response processes that protect users, contain harm, and strengthen your systems after a crisis.
• Integrate safety into your organization's culture with proven techniques for measurement, accountability, and persuasive communication.
• Utilize ready-to-use templates for resiliency plans, risk strategies, and checklists for brainstorming failures.

Stop reacting to disasters and start building systems that are designed to fail well.

Yonatan Zunger is Microsoft's CVP and Deputy CISO for AI and former CTO of Social at Google. Lea Kissner is the CISO of LinkedIn, the former CISO of Twitter, and a leading expert in privacy engineering and access control. Neil Coles is a Principal Security Engineer at Microsoft focused on developing methodologies for building safer AI systems. Juan Pablo Hernandez is a Principal Security Assurance Engineer Lead at Microsoft specializing in AI security and threat modeling. Harmony Mabrey is a product builder who developed the Harms Modeling framework at Microsoft. Phillip Misner is Head of AI Incident Detection & Response at Microsoft with extensive experience in cybersecurity crisis management.