Trustworthy and Resilient AI for Smart Energy Systems: Frontier AI, Reinforcement Learning, and Security

The energy sector is becoming a highly connected cyber-physical ecosystem in which distributed energy resources, electric vehicles, sensors, edge devices, market platforms, and operator interfaces interact continuously. AI and machine learning can improve forecasting, control, protection, and decision support, while reinforcement learning offers adaptive strategies for dynamic grid operation. At the same time, cyberattacks, data poisoning, false data injection, model misuse, privacy leakage, and unsafe autonomous actions introduce new risks. Large language models are emerging as a practical interface for power-system knowledge extraction, event interpretation, operational assistance, and security analysis. However, their reliability, safety, and governance in critical energy infrastructure remain unresolved.This Research Topic aims to advance secure, trustworthy, and deployable AI methods for smart energy systems by linking reinforcement learning, large language models, and cyber-physical security. We seek contributions that move beyond proof-of-concept algorithms toward validated frameworks for resilient grid operation, secure decision support, and human-in-the-loop automation. Particular attention will be given to methods that quantify risk, enforce safety constraints, improve interpretability, and withstand adversarial operating conditions. Recent advances in safe and robust RL, digital twins, graph learning, federated learning, retrieval-augmented generation, agentic LLM workflows, and cyber-resilience analytics create an opportunity to redesign how power-system data, models, standards, logs, and operator knowledge are integrated.We invite Original Research, Review, and Methods manuscripts. Topics include the following but are not limited to:- AI and reinforcement learning for power-system forecasting, operation, and control; - fault detection, diagnosis, and predictive maintenance for grids and electrical equipment; - reinforcement learning for wind farm control, demand response, voltage regulation, and distributed energy resource coordination; - AI-enabled detection and mitigation of cyber-physical attacks such as false data injection and denial-of-service; - data-driven stability assessment, contingency analysis, and resilience enhancement; - foundation-model-assisted applications for grid operation, fault analysis, and maintenance; - digital twins, benchmark systems, and real-time validation for smart grids, microgrids, EV charging, and virtual power plants. Topic editors: Guolong Liu, Gaoqi Liang, Huan Zhao, Huanxin Liao Source: https://www.frontiersin.org/research-topics/82016/trustworthy-and-resilient-ai-for-smart-energy-systems-frontier-ai-reinforcement-learning-and-security