PhishLens: An Interactive Phishing Email Training Platform with Customizable Email Generation

Phishing emails have long posed a threat to cyber security. They exploit malicious links or attachments to steal personal information and sensitive data, deliver malware and eventually cause significant financial loss. Although various measures have been deployed to identify phishing emails, attackers also constantly improve their techniques, increasing the ability gap between attackers and email service providers. Well-crafted phishing emails can easily break into recipients’ inboxes and deceive incautious users, especially those not adequately trained. To this end, we propose PhishLens, a Large Language Model (LLM) enabled phishing email training platform. It utilizes LLM to carefully create diverse phishing scenarios based on a rating module that evaluates rapidly evolved characteristics of phishing emails. It also provides real-time feedback and tracks user performance. Thus, the platform can provide personalized training based on user performance to improve the users’ ability to recognize phishing emails. To evaluate the effectiveness of PhishLens, we conduct real-world user studies to assess user performance and learning outcomes. Experimental results indicate that PhishLens significantly improves users’ ability to identify phishing emails.