Toward AI-pervasive electric energy systems: Concept, framework, vulnerability, and visions

Over the past decades, electric energy systems (EES) have undergone remarkable reforms in wide-area system interconnection, marketization, decarbonization, digitalization, and the electrification and integration of other sectors such as transportation, introducing unprecedented uncertainties and complexities into EES operation and management. To accommodate increasing changes while relaxing human burdens in daily routines, the latest trend is to integrate AI-based tools to complement traditional operation methods. Recent breakthroughs have enabled AI to handle nearly all kinds of operation tasks, with the potential to reshape EES operational and management frameworks. Given the existing successes of AI, a widespread integration of AI into EES all-sided daily routines becomes feasible, paving the way for advancing AI-driven EES in the future. AI methods have been discussed for their general shortcomings and risks. Particularly in EES studies, these discussions merely cover the capability boundaries and limitations of individual AI models. However, EES are vast, complex industrial systems and also safety-critical infrastructures, operated through massive interdependent and nested daily routines. When pervasively adopted, the concern of whether multiple AI applications can always ensure consistent and coherent interplays across varying complex operation scenarios arises. This implies that future EES may suffer from AI-induced vulnerabilities, particularly those arising from complex interactions among multiple AI applications, which have already exceeded the scope of traditional EES vulnerability theories and are still neither clearly understood nor systematically studied. This article seeks to articulate the concept and framework of AI-pervasive EES, pinpoint the potential vulnerabilities inherent to this future paradigm, propose a general taxonomy for their study, outline their underlying mechanisms and impacts, and suggest future directions. © 2026 The Author(s)