Ensuring safe and healthy food is a big challenge due to the complexity of food supply chains and their vulnerability to many internal and external factors. Recent research has shown that Artificial Intelligence (AI) based algorithms, in particularly data driven Bayesian Network (BN) models, are very suitable as a tool to predict future food safety risks and hence allowing food producers to take proper actions to avoid food safety problems to occur. Such models become even more powerful when data can be used from all actors in the supply chain (e.g. farmers, food producers, authorities) but data sharing is hampered by different interests, data security & data privacy. Federated learning (FL) may circumvent these problems as demonstrated in various areas of the life sciences mainly using linear models. In this research, a federated BN was developed for food fraud to demonstrate the potential of FL for the whole food safety domain. This concept consisted of three geographically different data stations hosting different sets of food fraud data which have been made FAIR (e.g. Findable, Accessible, Interoperable & Reusable). It is demonstrated that a BN model can be trained on the data of different data stations while the data never leaves its data station abiding security and sensitivity requirements and that this BN model performs similar to a BN model trained on the complete data set pulled into one data station. We demonstrated for the first time the applicability federated BN in food safety and anticipate that such concept may support stakeholders in the food supply chain for better decision making regarding food safety and food fraud control.