The increasing demand for trustworthy and explainable artificial intelligence (AI) has driven the adoption of techniques like fuzzy logic and federated learning (FL), especially in domains requiring transparency and reliability. Evolutionary Fuzzy Systems (EFS) are particularly relevant in this context, as they combine the adaptive learning of evolutionary algorithms with the interpretability of fuzzy logic, making them well-suited for eXplainable AI (XAI) and FL applications. However, a key challenge in deploying EFS in federated settings is that clients independently evolve fuzzy sets based on local data, leading to semantic inconsistencies in linguistic labels. This heterogeneity undermines the global model’s coherence and interpretability, affecting decision-making. To address this issue, we propose a synchronization mechanism that aligns fuzzy labels across clients. The approach involves centralizing fuzzy sets from all nodes, computing their average at a server, and redistributing a unified version to participants. This modification to standard EFS preserves data privacy while ensuring a consistent linguistic foundation, improving both robustness and explainability in federated EFS models.