Вдосконалення механізмів захисту корпоративної електронної пошти від фішингу на основі машинного навчання та протоколів SPF, DKIM і DMARC

Abstract

The paper considers an approach to improving mechanisms for protecting corporate email from phishing attacks based on the use of machine learning methods and the SPF, DKIM and DMARC authentication protocols. An analysis of modern threats associated with the spread of phishing messages, which are among the main causes of account compromise and leakage of confidential information in the corporate environment, is conducted. A model for detecting suspicious emails is proposed that combines the results of SPF, DKIM and DMARC checks with additional message characteristics, including header parameters, domain features and behavioral indicators of the sender. The use of machine learning algorithms makes it possible to automate message classification and increase the accuracy of phishing attack detection compared with traditional filtering methods. The expected result is a decrease in the number of successful phishing attacks, an increase in the level of information security of corporate email systems and an improvement in electronic correspondence monitoring processes.