This study investigates the effectiveness of diverse machine learning models in detecting phishing attempts, a critical concern exacerbated by the rise of remote work and telecommuting in the post-COVID era. Leveraging the capabilities of Machine Learning, particularly in phishing detection, we conducted experiments using both textual and numerical data extracted from URLs encompassing phishing and benign instances. Our findings reveal that a Convolutional Neural Network trained on textual data achieved an accuracy of approximately 91%. Conversely, utilizing numerical features from the same URLs, a Support Vector Machine and Multi-Layer Perceptron surpassed 99% accuracy, demonstrating superior performance across metrics such as precision, recall, and F1 score. Additionally, our analysis confirmed that these highly accurate models exhibited no signs of overfitting, indicating robust performance in phishing detection tasks.