<p>Highly sophisticated cyber-attack methods ranging from phishing to the likes thereof necessitate highly sophisticated detection techniques. In this enhanced security resilience system, the authors propose the use of a Zero Trust Threat Detection System with Spiking Neural Networks (SNNs). The process starts with data preprocessing using a one-hot encoded format for the Kaggle dataset to ensure a structured presentation of features. Recursive Feature Elimination is used for optimal feature extraction, whereas autoencoder methodologies are used for extracting vital patterns, fine-tuning input features for better classification accuracy. The Mean Squared Error loss function is used for training the autoencoder, which helps in minimizing error variance and maintaining important patterns with reduction of noise-induced distortions. Following feature processing, classification is done by an RMSprop-activated SNN that optimizes computation and learning dynamics. A key part of the framework is also the Zero-Trust Verification step, which fortifies security through assessment of classification outcomes and adaptive imposition of verification controls. By incorporating this phase, the detection framework avoids unauthorized intrusions and adapts to evolving threat landscapes. Experimental results demonstrate that the proposed model outperforms existing frameworks, including CNN-BiLSTM, SVM, and baseline SNN approaches, in detection quality and computational cost. The proposed Zero-Trust Threat Detection Scheme, Spiking Neural Network-based, achieves 99.99% accuracy, 99.11% precision, and 99.86% recall, outperforming conventional phishing detection models. The system is executed using Python, confirming the efficiency of machine learning-based cybersecurity tools and opening up a new paradigm for threat prevention.</p>