The fast growth of Internet technology has significantly changed online users’ experiences, while security concerns are becoming increasingly overpowering. Among these concerns, phishing stands out as a prominent criminal activity that uses social engineering and technology to steal a victim’s identification data and account information. According to the Anti-Phishing Working Group (APWG), the number of phishing detections increased by 46 in the first quarter of 2018 compared to the fourth quarter of 2017. So to overcome these situations below paper introduces a phishing detection system using a hybrid machine learning approach based on URL attributes. It addresses the growing threat of phishing attacks that exploit email manipulation and fake websites to deceive users and steal sensitive data. The study employs a phishing URL dataset with over 11,000 websites, extracted from a reputable repository. After pre-processing, a hybrid machine learning model, which includes Decision Tree, Random Forest, and XGB is employed to safeguard against phishing URLs. The proposed approach undergoes evaluation with key metrics such as precision, accuracy, recall, F1-score, and specificity. Results demonstrate that the proposed method surpasses other models, achieving superior accuracy and efficiency in detecting phishing attacks.