Representing, reasoning and predicting fraud using fraud plans

The efforts of fraudsters to think up new ways of committing fraud, and of law enforcers to detect and prosecute those fraud, often feel like a long-running competition. Yet all too often, law enforcement is accused of falling far behind the fraudsters, especially in situations where the responsibility for detecting frauds falls on non-specialists in security or on the general public. This paper presents a format to help understand how frauds work. The paper has three key messages: firstly, that every type of fraud plan can be represented as a specialisation of a generic fraud plan; secondly, that every type of fraud has `red flags' that ought to make potential victims suspicious enough to check whether a particular transaction might be fraudulent; and thirdly, that these `red flags' can be linked to the fraud plans because they arise as a natural consequence of the steps in the fraud plan. Several different types of fraud are described and generic fraud plans, specialised fraud plans and `red flags' are presented. The paper concludes by showing how a particular fraud plan could have been used to predict a new fraud that arose in 2010, and speculates on future frauds that various fraud plans might predict.