Philleted Phish – detecting phishing emails using sentiment analysis

Grant awarded: £31,000

PI: Trevor Wood, supported by Iryna Yevseyeva, Vitor Basto Fernandes, and Eerke Boiten

Funder: CyberASAP, Innovate UK

It is estimated that cyber-attacks cost UK businesses £37bn per year, most of which ends up in the hands of organised crime. It is also estimated that around 90% of these attacks start with a phishing email.

Many phishing emails have been relatively easy to spot in the past as they contain poor spelling, grammar, and sentence structure. However, with generative AI now being more accessible, it is easier to make phishing emails more realistic and remove these obvious markers. Moreover, these phishing emails are designed to bypass spam filters without being detected.

Philleted Phish will detect phishing emails, whether they attempt to trick victims into visiting a phishing website or perform some other action, such as transferring money to the attacker’s bank account or providing information useful to a larger social engineering attack. Unlike current phishing email detection software, it can be used by domestic users, micro-businesses, and SMEs without requiring expensive and specialised hardware.