Abstract

Phishing is the most prominent cyber-crime that uses camouflaged e-mail as a weapon. In simple words, it is defined as the strategy adopted by fraudsters in order to get private details from persons by professing to be from well-known channels like offices, banks, or government organizations. In this era of modernization, electronic mail is widely used as a communication channel for both private and professional purposes. The particulars exchanged over emails are often confidential and sensitive, for example bank statements, payment bills, debit-credit reports, and authentication data. This makes e-mails precious for hackers because they can exploit these details for malicious intent. The main goal of the attackers is to acquire personal details by deceiving the e-mail recipient into clicking a malicious link or downloading an attachment under false pretences. In the last few years, there has been an exponential rise in cyber threats, and phishing e-mails in particular have resulted in huge monetary and identity losses. Several models have been developed to separate ham and phished e-mails, but attackers are always trying new methods to invade the privacy of the people. Spam emails are still a major cybersecurity risk because they can be used to steal sensitive data from unwary users. This study investigates using machine learning to analyze email content and identify phishing attempts. To categorize emails as valid (“ham”) or phishing-related (“spam”), we preprocess the text, use natural language processing techniques, and deploy an LSTM neural network on a dataset of labeled emails. The high accuracy of the suggested model shows that it has the potential to be a useful tool for email security.