Not a week goes by when we don't see a newspaper article covering the latest large scale incident of fraud. And increasingly these are related to digital banking. According to ThreatMetrix new cybercrime insights, in the first half of 2018 mobile fraud attack rates across the globe rose 24%, when compared to the first half of 2017. Put in real terms, financial institutions were besieged with 81 million cybercrime attacks in the first six months and of these, 27 million were targeting mobile banking channels.
The impact of this increase to banks is colossal; falling share prices, fines, reputational and client/partner relationship damage etc. It also means more pressure on systems and processes to accurately identify these. This means increased operational costs and more time spent on identifying financial crime. In this developing digital environment, where we live in a real-time world, fighting sophisticated financial crime techniques feels like an impossible challenge. But it is possible to accurately outsmart these criminals and maintain efficiency. How?
Digital demands real-time fraud analysis
Bank customers expect a banking experience that matches their digital lifestyles. In fact by 2021, one out of every two adults in the world will user a smartphone, tablet, PC or smartwatch to access financial services. These digital users are your customers. They expect to be able to transact instantly and this means that fraud needs to be identified immediately. But real-time transactions however make identifying illegal transactions a real challenge. This is because they are processed as they happen and cannot be reversed – there is no time for manual fraud review steps. Addressing fraud in real-time using just historical data alone isn't sufficient. Artificial Intelligence (AI) may hold the answer.
Accuracy through AI
One of AI's key features is its ability to detect patterns and recognise small deviations that occur which seem irrelevant to create an intricate profile. This functionality is perfect to support financial crime mitigation. For example, location data can be collected from a customer's phone on an ongoing basis; looking beyond just a one-time snapshot of a mobile payment transaction. This machine learning approach would enable a profile of habitual behaviour for that customers mobile payments activity. As a result the profile is more accurate with alert detection and false positives are reduced.
As well as increased profile accuracy, machine learning could also offer greater efficiency (and accuracy) when investigating possible fraudulent transactions. Many existing fraud systems have no automated decision workflow and rely mostly on manual review. This makes banks vulnerable to exploitation, e.g. hitting banks with a number of attacks in short timeframes. This type of attack bombards the system, making it hard for a banks fraud department to address the multitude of cases and forcing them to act quickly to both protect their customers and themselves. As machine learning capabilities develop, there is no reason why the current roles of human employees investigating these transactions couldn't be managed using this new capability, making a bank's business operations more efficient. In addition, because manual investigation takes time (many bank policies as part of the investigation process may take 45 to 60 days to refund customers who have been defrauded), this often means a poor customer experience. If the use of AI can shorten this period then trust is increased and damage can be limited.
Reducing false positives
False positives are a major issue for bank operations in general. From talking to banks with standard financial crime software, between 5% to 7% of transactions are identified in error as related to a financial crime ie. money laundering, fraud, sanctions breach etc. That's a lot of transactions to review (often manually). Potentially good customers may be treated unfairly unless they are identified as 'false positives' quickly and addressed effectively.
It is possible to get a false positive rate down however. Significantly reduced false positive incidents can be realised by using sophisticated algorithms and highly effective scanning methods. This combination of algorithms and methods allows for a higher threshold in fuzzy matching and, as a consequence, reduces the number of false positives without compromising accuracy in hit detection, and ultimately realising enhanced efficiencies. Low false positive rates mean seamless transactions, and in today's digital work, seamlessness is key.
Achieving real-time accuracy and efficiency
In PWC's 2018 survey, 82% of respondents agree that the use of continuous real-time monitoring assists their organisation in combating economic crime. Systems that use AI and real-time behaviour analysis build user and customer profiles to detect and stop suspicious transactions with elements such as unusual amounts, abnormal frequency, suspicious location and transactions to not‑seen‑before business partners. They often take into account elements such as transaction amount, currency, transaction type and frequency parameters, which can be combined and compared to 'usual behaviour' or predefined patterns.
Ultimately, in the fight against financial crime, success all comes down to automation. Working with an agile, real-time system that utilises AI and sophisticated algorithms with advanced scanning methods provide banks with an essential function; the ability to accurately stop suspicious transactions before the funds are moved, outsmarting the criminals efficiently, allowing banks to focus on what's important – their customers.
Written by Marlène Meli, product manager, FCM - Temenos