Examination of a new defense mechanism: Honeywords

Abstract

Past experiences show us that password breach is still one of the main methods of attackers to obtain personal or sensitive user data. Basically, assuming they have access to list of hashed passwords, they apply guessing attacks, i.e., attempt to guess a password by trying a large number of possibilities. We certainly need to change our way of thinking and use a novel and creative approach in order to protect our passwords. In fact, there are already novel attempts to provide password protection. The Honeywords system of Juels and Rivest is one of them which provides a detection mechanism for password breaches. Roughly speaking, they propose a method for password-based authentication systems where fake passwords, i.e., “honeywords” are added into a password file, in order to detect impersonation. Their solution includes an auxiliary secure server called “honeychecker” which can distinguish a user’s real password among her honeywords and immediately sets off an alarm whenever a honeyword is used. However, they also pointed out that their system needs to be improved in various ways by highlighting some open problems. In this paper, after revisiting the security of their proposal, we specifically focus on and aim to solve a highlighted open problem, i.e., active attacks where the adversary modifies the code running on either the login server or the honeychecker.