Diabetes mellitus is one of the most global health challenges, needing early detection and effective management to reduce long-term issues. Artificial intelligence (AI) techniques have shown promising capabilities in predicting diabetes using structured medical data, but most studies have focused on improving predictive accuracy while neglecting cybersecurity and patients data protection. This study suggests an integrated framework that combines machine learning with cybersecurity solutions to build a smart diabetic care system that is both accurate and secure. The proposed system applies advanced encryption, anomaly detection, and differential privacy techniques to protect sensitive health data while maintaining high diagnostic performance. Experimental results demonstrate that the integrated approach achieves superior predictive accuracy compared to conventional models, while significantly enhancing resistance to common cyber threats.