This paper proposes a comparative spam email detection framework that integrates traditional machine learning and deep learning models with a blockchain-supported auditability layer to enhance digital security, transparency, and trustworthiness. The study evaluates six representative classification approaches, including Naive Bayes, Support Vector Machine, Random Forest, Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), and Bidirectional Encoder Representations from Transformers (BERT). A unified experimental workflow is implemented consisting of data preprocessing, feature extraction, model training, and comparative evaluation using Accuracy, Precision, Recall, F1-score, and ROC-AUC metrics. In addition to classification performance analysis, the framework incorporates a lightweight blockchain-based verification mechanism that records classification metadata and cryptographic hashes to ensure integrity and traceability of spam filtering results. Experimental findings demonstrate that deep learning models, particularly BERT and LSTM, achieve superior contextual understanding and higher detection accuracy, while traditional machine learning methods provide lower computational complexity and faster execution suitable for lightweight environments. The proposed framework contributes a reproducible benchmarking methodology for intelligent spam detection and demonstrates how blockchain-supported auditability can improve transparency and reliability in AI-driven cybersecurity systems.