Exposing Phishing Strategies: Leveraging Machine Learning To Examine URL Characteristics For Proactive Threat Identification

The extensive integration of digital technologies has significantly transformed our lives, ushering in an era where almost all human activities are interconnected with the internet. However, this digital landscape, commonly known as cyberspace, has evolved into a breeding ground for adversaries who exploit unsuspecting internet users relentlessly. The imperative to make technology accessible has unintentionally allowed even those with limited technical knowledge to leverage sophisticated attack techniques devised by adversaries. Amid the rising prevalence of cyber adversaries and their formidable arsenal, one of the most concerning threats is phishing attacks. These attacks have reached alarming levels of success, highlighting the urgent necessity for effective countermeasures. While conventional defense strategies involve creating lists, like blacklists or whitelists, to label safe websites, this approach falls short in countering zero-hour (0-h) attacks. The swift pace of such attacks, often facilitated by phishing-as-a-service (PhaaS) technologies, creates a critical gap between website classification and list updates, rendering users vulnerable during this interval. To tackle this critical challenge, there is a compelling requirement for proactive systems capable of preemptively detecting and mitigating phishing attacks. This project introduces an approach that addresses the limitations of traditional lists and their inability to combat 0-h attacks. By presenting an innovative method for detecting phishing websites, this research contributes to bolstering internet security, safeguarding users from adversaries and their changing tactics as they navigate the intricate cyber landscape.