Displacing big data: How criminals cheat the system

Abstract

Abstract: Many technical approaches for detecting and preventing cy-bercrime utilise big data and machine learning, drawing upon knowledgeabout the behaviour of legitimate customers and indicators of cyber-crime. These include fraud detection systems, behavioural analysis, spamdetection, intrusion detection systems, anti-virus software, and denial ofservice attack protection. However, criminals have adapted their meth-ods in response to big data systems. We present case studies for a numberof different cybercrime types to highlight the methods used for cheatingsuch systems. We argue that big data solutions are not a silver bulletapproach to disrupting cybercrime, but rather represent a Red Queen'srace, requiring constant running to stay in one spot.