The increasing demand for raw materials and shorter product cycles have driven the advancement of the circular economy, with the European Union focusing on reducing raw material consumption and ensuring the circularity of products and components. Remanufacturing, a key element of circular economy, faces challenges in assessing the condition of products and components, a process traditionally carried out manually, requiring expert knowledge. This paper aims to address the lack of diverse datasets by generating synthetic datasets to enable AI-supported visual inspections in remanufacturing. Specifically, this paper focuses on generating realistic wear patterns, in the example of corrosion effects, with starter motors as the reference product. The research explores different diffusion models, emphasizing globally generated corrosion patterns. The outcomes include the creation of a synthetic dataset, the development of an accurate wear classification model, and a validated framework for integrating AI-based inspections in remanufacturing operations. This approach aims to improve efficiency and quality in remanufacturing processes, contributing to a circular economy.