The inclusion of machine learning (ML) into outcome-based education (OBE) has transformed academic assessment, course delivery, and the creation of individualized learning paths. Learning management systems based on machine learning enable teachers to tailor their teaching strategies to align with learning objectives, leveraging data-driven insights into student performance and the effectiveness of the curriculum. Conversely, the use of these technologies raises several significant ethical questions, particularly regarding data privacy, informed consent, algorithmic bias, and transparency in decision-making. These challenges draw attention to a substantial issue: especially when student data is aggregated or processed in centralized architectures, the current machine learning-based OBE models do not adequately guarantee data privacy, model fairness, or ethical compliance. Improper handling of this data could result in student privacy breaches or unfair evaluations. Addressing these problems builds confidence and ensures continuous GDPR and FERPA privacy rule compliance. This article presents a Privacy-Preserving Ethical Learning Model (PPELM) designed specifically for machine learning-driven OBE systems, which is proposed in the present work as a means of circumventing these constraints. Results of a comparison with centralized machine learning models indicate that PPELM performs better with less data exposure and more fairness across a wide range of student groups.