The increasing demand of e-commerce and online retail services in the past years has required development of strong security protocols able of protecting the privacy and safety of consumers. Machine learning has proven an essential tool in the analysis and reaction to cybersecurity threats, thus ensuring the security and privacy of users of online retail stores all over the globe. In this sense, machine learning protocols have proven essential in the prevention of cyberattacks like cross site scripting (XSS), in the identification of potential threats affecting the financial results and overall performance of a company, or the surveyance of the behavior of the individuals within an online community to identify excessively authoritative and potentially harmful members within different communities. The present paper analyzes the state-of-the-art advances in this field, providing an overview of the different methods that have proven their validity to develop machine learning-based models to ensure the security and privacy of users of online retail stores. The proposed machine learning methods use validation protocols such as NLP, SVM-based algorithms, neural networks, and text mining. Based on such analysis, the paper suggests potential approaches to a better implementation of machine learning in cybersecurity, enabling a faster and more effective response against the threat.