Phishing means cybercrime where attackers pose themselves as a known legitimate user to plifer sensitive information such as passwords and financial details through misleading emails or websites. Hence, this type of continuous threat persists despite constant improvement in countermeasures because this type of sophisticated phishing becomes even more complex to avoid. Machine learning techniques have proved useful in detecting phishing attempts by detecting patterns that indicate malicious activity. The survey summarizes about 25 research papers, categorizing the detection methods based on ML techniques used, datasets, and performance metrics adopted like accuracy and F1 score. This review includes algorithms of many classification techniques, such as Naive Bayes, Random Forest, SVM, and CNN. Through the above research questions, some of the most common techniques and the datasets that are found in the existing literature along with which algorithms show the best performance for accuracy are outlined.