To deploy AI safely, organizations should understand potential risks and create a comprehensive management plan that includes the entire system and its human components. Key principles involve continuous risk assessment, creating a written safety plan, and adapting strategies to address both conventional and AI-specific challenges, such as inherent errors in AI systems. Practical guidance and resources for safe AI deployment are also highlighted in the ongoing series from Microsoft’s Deputy CISOs.