Accelerating incident response using generative AI

This report describes Google’s use of generative AI to accelerate and improve security incident communications by structuring inputs with tags, iterating prompts with exemplars, and integrating a human-in-the-loop drafting tool; measured outcomes include ~51% time savings for incident summaries and ~53% for executive updates, with safeguards to prevent data retention and mitigate hallucinations (e.g., minimum input size checks).