Low-risk instant payments - the approach of the fraud detection solution matters most

New payment options and greater convenience led to large volumes of data to be analysed in real-time; as a result, Roy Prayikulam, INFORM agrees that the approach of the fraud detection solution matters the most

With the introduction of the SEPA Instant Payment procedure, banks have been able to significantly increase the user-friendliness of their transfer procedures. However, the service is associated with an increased risk of fraud. Fraud detection solutions can help minimise this risk. But not all solutions are the same. Different approaches are being taken. And not all of them can keep up with the constantly evolving attack patterns of fraudsters.

Since 2017, SEPA Instant Payment has been on the rise as a payment method in Germany and Europe. Almost all German banks can now offer their customers real-time transfers. Some institutions are even using the procedure for granting loans. A financial transaction that used to take days to process can now be completed in a matter of seconds thanks to instant payments. For customers, using the service is as easy as can be. All they have to do is log in via their online or mobile banking account, authenticate via their TAN and - as usual - authorise the transaction. However, the process is not without additional risk for banks and customers alike.

Now, the European Payment Initiative (EPI) is to demand a clear commitment to electronic payments from its current 31 participating banks. The new procedure is based on SEPA Instant Payments and is intended to promote digital payment processing in Europe. However, the fact that secure fraud detection solutions are currently more relevant than ever is not only due to the increasing number of banks that want to offer instant payments. In 2019, for example, the number of faster payment fraud-related losses in the UK increased by 33 percent to GBP 330 million, according to UKFinance, with 98 percent of these losses having to be borne by the banking industry. Due to the pandemic and the associated increased digital payments via online banking and online shopping, these numbers and sums have been rising ever since.

Fraud risks with instant payments

The market continues to offer consumers new options and greater convenience. At the same time, the amount of data to be processed in a short time is growing exponentially. New players, payment systems and processes - keyword instant payment - require fraud prevention systems to reliably process ever larger transaction volumes in real time. This is because the high speed of order processing means there is little time left to check for potential fraud. A real problem. Bank accounts have been at the top of the list of targets for cybercriminals for years. Using phishing and spear phishing attacks, they obtain information about their victims, which they then use to either gain direct access to their accounts or manipulate them into voluntarily making transfers to the fraudsters' accounts.

Real-time monitoring as a solution

To detect these and other types of fraud without slowing down the transaction speed and, if possible, to stop them without losing money, the majority of banks now use automated fraud detection solutions. These monitor user behaviour continuously and in real time and analyse it for anomalies. Machine learning methods are increasingly being used for this purpose. However, many of these solutions have a problem: their success is essentially based on the maturity of the machine learning models they use. These must be intensively trained before they can be used for the first time with a reasonable hit rate: Time that is not available to banks.

With RiskShield, we have taken a different approach. RiskShield also uses machine learning - but its fraud detection method is based on a hybrid artificial intelligence approach. It combines data-driven machine learning methods with knowledge-driven methods, such as fuzzy logic-based scorecards and watch lists. This enables such a fraud detection solution via real-time risk assessment and monitoring to draw the right conclusions even when data is imprecise. Thus, the solution can be effectively deployed practically from the very beginning.

The coming danger - multi-channel fraud

The hybrid AI approach with its intelligent combination of technologies is therefore best equipped to deal with more complex fraud scenarios. Cybercriminals and fraudsters have long since seen through the approach of common real-time monitoring solutions and have begun to adapt their attack vectors. Instead of running their attacks through just one channel - mobile banking, for example - they are increasingly splitting them across multiple channels. For example, when they gain access to a bank account, they don't just empty it. Instead, they also secure additional information, for example about their victim's credit cards, which they then use to make purchases on the Internet at the victim's expense. Since the analyses of most fraud detection solutions focus on user behaviour on one channel, they are unable to detect such cross-channel fraud. Although they are capable of recording and analysing user behaviour on different channels - and are thus multi-channel capable - they are unable to perform linked data analysis. In such cases, only cross-channel-capable solutions can help. With these, all data on user behaviour, from online and mobile banking to cash withdrawals at ATMs, SEPA transfers and contacts with customer service, to ecommerce transactions, can be examined in a coherent manner. More importantly, such and other behavioural data of a user can be combined even if they originate from different payment service providers. Even data from regulators, such as watch lists, can be dynamically integrated into the overall analysis of user behaviour.

As fraud prevention has access to more and more data from a wide variety of sources, and as fraudsters' attack vectors become more sophisticated, it is likely that fraud detection solutions in the future - if they are to continue to deliver effective results - will not be able to do without such cross-channel functionality. What is clear is that instant payments are becoming essential to more and more sectors of society through initiatives such as EPI. Effective fraud prevention solutions are needed to protect banks and customers from fraud-related losses.

About Roy Prayikulam

Roy Prayikulam is SVP Risk & Fraud division at Inform, an internationally active optimisation specialist based in Aachen, Germany. After graduating from RWTH Aachen University, he began his career in the field of financial crime prevention in 2009 at Inform, where he has since held various roles, from Business Consultant to Project Manager to Head of Professional Services. Roy Prayikulam has extensive experience in complex IT integration projects for the financial sector, for example in the areas of acquiring, card issuing, internet banking and AML compliance.

About INFORM

INFORM develops software for the optimisation of business processes using Digital Decision Making based on Artificial Intelligence and Operations Research. The company supplements classic IT systems and increases the profitability and resilience of many companies. As a leader in providing intelligent, customer-centric fraud prevention and AML compliance solutions, with RiskShield INFORM offers a multi-channel platform that detects and manages suspicious activities, minimizing losses and optimising efficiencies using advanced analytics, machine learning and intuitive rule management controls. RiskShield provides a robust solution with proven fraud detection results that are reliable, fast, and responsive. Today, more than 800 software engineers, data analysts and consultants serve more than 1,000 customers worldwide from container terminals, commercial airports, financial services providers, manufacturing companies and wholesalers to warehousing and handling centres as well as transport companies.

www.riskshield.com | video: http://bit.do/riskshield_video