Power systems are transforming into digitalized smart grids where cyber, physical, and social domains converge. In parallel, Generative AI (GenAI) offers new capabilities to address these challenges. While prior research explores AI in specific power sector applications, comprehensive assessments of GenAI’s broader impacts remain limited. This study introduces a three-dimensional framework encompassing operational value, safety and trustworthiness, and regulatory implications. We propose a structured methodology for building GenAI use-cases and demonstrate its effectiveness through a predictive maintenance example, achieving over 31% performance improvement compared to traditional AI. Additionally, we assess GenAI safety risks, present a taxonomy with measurable trust metrics, and offer policy recommendations aligned with global practices. This work supports the responsible integration of GenAI into power sector operations, balancing innovation and value creation with safety, reliability, and regulatory readiness.
