Using AI in financial fraud detection can help banks save unnecessary losses

According to latest reports by McAfee, cyber crime and financial fraud are currently costing the global economy $600 billion. This equates to 0.8 percent of global GDP and necessitates the need for stronger than ever safety mechanisms. Established financial behemoths are leading from the front here, by integrating advanced cyber-security solutions, to protect their data and user privacy. Smaller players are following suit in an industry that is worth more than $153 billion globally. As the financial services landscape changes, technology is providing the tools and platforms for rapid transformation. This evolution is being strengthened by **Big Data** , which has become deeper and broader than ever before. Banks and financial institutions now have access to more information about customers and their spending habits. While this has led to the deployment of chat bots for answering customer queries, banks are now more interested in applying innovative technologies such as, AI and machine learning to mitigate financial risks. PwC’s 2018 Global Economic Crime and Fraud Survey states that 49 percent of global organisations have experienced financial crime in the past two years. As global transactions over digital channels have increased, so has the possibility of electronic fraud. Additionally, cyber criminals now also have the newfound ability to leverage cryptocurrency exchanges to transfer and withdraw stolen money from their fraudulent activities. Such unregulated channels make it imperative for banks and financial institutions to deploy advanced techniques to fight cyber crime. [caption id=“attachment_4654591” align=“alignnone” width=“1280”] Representational Image[/caption] The role of AI in improving the detection of financial fraud When it comes to financial risk mitigation, fraud detection in real-time goes a long way towards improving customer experiences and boosting company reputation. This is why banks and financial institutions are turning to modern solutions and advanced data models offered by AI and machine learning. By dynamically conducting fund flow analytics in real-time, such solutions can effectively pinpoint fraudulent transactions. Moreover, they can also reduce the possibility of false positives (situations where real transactions are treated as frauds, transactions are declined, and accounts are suspended) and false negatives (situations where real threats are missed). The flexibility offered by AI and advanced analytics can now be applied across all banking functions — right from customer engagement to improved risk management. Mastercard was among the first financial organisations to deploy such solutions for fraud management, and it was thus able to reduce the rate of false declines its customers faced by a whopping 80 percent. Previously, the organisation had to deal with several instances of real transactions getting flagged or real threats getting missed as the rules for classifying millions of transactions could not be consistently applied. By subsequently integrating AI into the validation of these transactions, the company has changed its fortunes completely. Thanks to the availability of large volumes of personalised customer data and transnational history, AI and machine learning can now be used to instantly identify customer behaviour patterns that are out of the ordinary. This can help establish a massive database of information about specific customer patterns, which can then be leveraged to upsell new products or services. But most importantly, this repository of data can provide the possibility of early detection of any abnormal behaviour that may lead to cases of financial fraud or theft. Transforming the financial services industry – one cluster at a time Ironically, advancements in AI technology are also leading to a rise in AI-enabled cyber attacks as banks and financial institutions are storing data on private/public infrastructure. These institutions should thus focus on the implementation of learning models to identify user fingerprints and place a set of similar users into clusters. Fraud detection applications routinely review customers’ social media, work history, education and more, to ascertain if an individual’s financial activities are in sync with those of the cluster they belong to. Such sophisticated models can be continuously updated to include a wide variety of changing customer data to automatically adjust what constitutes financial fraud. At present, many banks and financial institutions use a two-layered detection process to identify the possibility of financial fraud. The first screening stage is undertaken by AI, but the second stage involves manual checking — a situation that is still vulnerable to the possibility of human error or tampering. These institutions must work towards a scenario wherein this second stage can be completely removed. Financial fraud has been a constant throughout human history, and advancements in technology have made it more complex and difficult to contain. However, banks and financial institutions now also have the ability to leverage self-learning technology to identify such activities and prevent them. Not only will this reduce the financial burden of cyber crime, it will also improve their reputation and customer loyalty. AI is already causing massive upheaval in the banking and financial services industry, and a cyber crime-free future looks increasingly within reach. The author is president and head, Banking, Financial Services & Insurance (BFSI), Healthcare and Life Science at Infosys