This research paper addresses the critical need for robust email security measures in the face of evolving cyber threats, specifically focusing on innovative approaches to phishing detection and prevention. Initially, traditional machine learning techniques such as support vector machine (SVM), gradient boosting, multinomial Naive Bayes, and random forest are employed to analyze a dataset of email texts. While these methods demonstrate promise, the research takes a transformative step by integrating T5, a cutting-edge transformer-based model. T5 not only outperforms traditional algorithms but also showcases remarkable adaptability in discerning intricate patterns within email content. A comparative analysis of classifiers reveals the superior accuracy of transformer-based models, marking a paradigm shift in email security. The study acknowledges limitations and considerations, including dataset biases and nuanced implications of false positives and false negatives. In the subsequent phase, the research explores fine-tuning T5 for phishing email detection, demonstrating robust performance during training and commendable accuracy during evaluation. Practical applications are realized through the implementation of a Gradio front-end interface. This research contributes not only to academic discourse on email security but also offers tangible insights for deploying AI-enhanced models in real-world scenarios. The integration of T5 underscores its transformative potential in fortifying email security against dynamic phishing threats.