“In the rush to contain COVID disruption, public and private sector organisations implemented measures, at pace, to protect the economy. Expediency may compromise some aspects of the usual safeguards to design out fraud, leaving systems open to unanticipated exploitation by (cyber)criminals. In reality, any significant change or new initiative will attract fraudsters, even if designed and implemented well. The quick roll-out of the Government’s Bounce Back Loan scheme is just one example of this.
“The pandemic and the accompanying drive to digital channels has made it easier for organised crime groups and opportunistic individuals to exploit the situation. With new initiatives like this, there will inevitably be error as well as fraud, which can make it more difficult to pinpoint the genuine fraud cases. Ultimately it’s law-abiding businesses and taxpayers that pay the price as these groups steal from the public purse or operate under the guise of actual government bodies to trick the less-wary citizen.
“The ‘unprecedented’ nature of the pandemic also hampers investigators and automated systems alike as components such as Anomaly Detection, for example, which relies on recognising significant deviations from normal behaviour, is more difficult given the raft of new initiatives and changes in what is the new “normal” behaviour brought on by the pandemic. Nevertheless automated anti-fraud technology using AI is one key to catching the criminals and helping to protect the public.
“In Banking, fraud detection models must quickly adapt to the new loan systems necessitated by COVID, for example by making use of scenario-based modelling to quickly learn from multiple ‘what-if’ scenarios and understand all the key inputs in this new environment. From here, efficient models with a low rate of false positives can help detect criminal behaviour and protect public funds at this critical time.
“Even the human isolation that has been the nature of the pandemic aids the criminals. Humans are very good at sharing concerns and networking starts to build a group picture that reinforces legitimate doubts. When this is disrupted it gives the crook more time to operate. Luckily, many modern systems incorporate social network analysis that allows the group picture to be assessed by the models that spot fraud.
“Being wary and asking yourself constantly whether it could be fraud remains a top tip and its always best to talk if in doubt.”
Simon Dennis, Future Government and AI Evangelist, SAS UK