Abstract: In today’s business landscape, phone conversations between customers and representatives yield a wealth of information, ranging from personal details to generic topics. While such insights are invaluable for businesses, the advent of legislation like GDPR necessitates careful handling of personal information. This paper explores the application of two machine learning algorithms to classify and safeguard sensitive data extracted from transcribed phone calls. Leveraging Naive Bayes and Support Vector Machine algorithms, we employ an iterative system development method to build a classification model. Through rigorous evaluation techniques including 10-fold cross-validation, learning curves, classification reports, and ROC curves, we assess the efficacy of the system. Our findings indicate that higher dataset volumes contribute to increased accuracy, with preprocessing techniques further enhancing model performance. This research underscores the potential of machine learning in fortifying customer privacy while optimizing service quality within organizations.