User Interfaces and Algorithms for Fighting Phishing

Google Tech Talks January 4, 2007 ABSTRACT In this talk, I will discuss our research group's latest work in developing user interfaces and algorithms for anti-phishing. With respect to user interfaces, I will describe the effectiveness of various interventions we have been developing for training people not to fall for attacks. With respect to algorithms, I will describe PILFER, a machine learning approach for finding phishing attacks in email with 99.8% accuracy, as well as CANTINA, a content-based approach to detecting phishing web pages using the Google search engine itself. A comparative study shows that CANTINA can find ~95% of phishing web pages. Credits: Speaker:Jason Hong