Phishing is one of the most dangerous threats in online security since attackers lie to people and obtain their confidential information by creating fake web pages that are nearly the same as real ones. To address this, we have developed an advanced system that combines machine learning with natural language processing will be designed to better detect phishing websites and enhance the communication relationship between the users and the website. We analyzed a dataset of 10,000 websites, where half were phishing websites and the other half were legitimate, in order to extract fundamental features that primarily include the structures of URLs, domain characteristics, and website behaviors. In this context, XG-Boost was found to be the best among the different models that were run, with an accuracy of 97% for detecting phishing attempts. To further enhance user experience, we have integrated a multilingual chatbot created using Dialog—flow, designed to engage users in the English, Hindi, and Marathi languages. It is because of integrating accurate machine learning capabilities with an active multilingual chatbot that our system provides the strongest and friendliest solution to combat phishing attacks.