Guest Editorial Introduction to the Special Issue on Responsible and Federated Foundation Models for Industrial IoT

We are witnessing unprecedented progress in foundation models. Foundation models have fundamentally altered our understanding of what AI can achieve. The Industrial Internet of Things (IIoT) faces both great opportunities and complex challenges. Foundation models offer good potential for unified intelligence that can adapt to diverse manufacturing processes, predictive maintenance scenarios, quality control systems, and supply chain optimization challenges. This special issue of IEEE INTERNET OF THINGS addresses a critical question: how can we harness the transformative power of foundation models while ensuring responsible deployment, data privacy, and collaborative efficiency in an industrial environment? Foundation models, with their remarkable capabilities to generalize across domains and tasks, present new opportunities for IIoT applications. However, the industrial context presents unique constraints, including real-time responsiveness, safetycritical reliability, and protection of proprietary operational data. The integration of federated learning with foundation models represents a promising shift toward collaborative intelligence by respecting data sovereignty at the same time. In industrial settings, where operational data often contain trade secrets, competitive advantages, and sensitive process parameters, the ability to train sophisticated models without centralizing data is essential.