Democratization of machine learning, behavioral biometrics, payments intelligence hubs and authorized push payment fraud are some trends to monitor.
As we head into the new decade, it is imperative that the leaders of banks, fintechs, as well as other financial institutions across Asia keep an eye on the rapidly evolving fraud landscape to ensure they stay one step ahead of fraudsters. Here are five of the major trends that I foresee rising to prominence in 2020, that will likely play a large role in the evolution of fraud prevention in the early parts of the next decade.
1. Machine Learning variants in fraud prevention
Machine learning has been a big area of interest for banks and financial institutions over the past few years, but a trend is emerging, and that is to take a more ‘democratized’ approach to machine learning: Making machine learning models more accessible and enabling a step-wise approach to their integration.
Fraud experts can tap into their existing expertise and knowledge, inputting into democratized machine learning models the specific correlations between transaction types that are strong indicators of fraud.
With a more democratized approach to machine learning, financial crime and risk managers with specific subject matter expertise are equipped with modeling tools that they can build, test and deploy quickly on their own to address specific threats. This also helps fraud experts explain their choice of features, their use of model scores and the actions taken, to management, auditors and regulators as required.
More financial institutions will adopt this approach in 2020, as it will allow them to quickly show the Return On Investments in their machine learning and fraud prevention initiatives.
We will also see the rise of unsupervised machine learning, which allows the machine to be trained with data that has not been classified or labelled; this allows such machine learning systems to find novel correlations in an automated way without the unconscious bias that humans may unwittingly apply when analyzing trends. As such, this branch of machine learning will likely become a must-have for enterprise fraud protection.
2. User behavior analytics for authentication
We will continue to see advancements in user behaviour analytics (UBA) as organizations recognize its value in continuous risk or authentication assessment. Organizations can use this to assess user behaviour; whether it’s a swipe on the phone, a wiggle of the mouse or a tap on the keyboard, making it quicker and easier to detect unusual or suspicious behaviour.
Behavioral biometrics is a particularly promising area of innovation, and companies such as BioCatch are continuously bringing new methods of fraud detection to the table, helping the industry to detect subtle hesitations that may indicate the payer is being socially engineered, coerced or otherwise manipulated as we are seeing with Authorized Push Payment (APP) fraud in the immediate payment channel.
3. The development of Payments Intelligence Hubs
Another potential trend to emerge in the new year is the shift towards centralized payments intelligence hubs, as enterprise-wide fraud management becomes too narrow an approach to risk management.
With the formation of such hubs, information around fraudulent activity can be shared via a central infrastructure or similar networks to improve fraud detection and prevention. For example, if one bank gets hit by fraud exhibiting a particular pattern, other banks could learn from it and adapt their fraud models accordingly. This creates a win-win situation between banks, payment networks and ultimately, the end customers.
4. Application fraud will continue to grow
I foresee the rise of application fraud continuing into 2020. Across the world, identity scans are largely broken, meaning that synthetic IDs and pure identity theft will continue to increase, especially as banks and credit grantors continue to neglect reporting these losses.
It also looks likely that fraudsters will continue their attacks on central infrastructures that manage digital address books for immediate payment accounts, with one APAC example being the recent Australian PayID attacks. As such, financial institutions and financial service providers will need to be vigilant to stay one step ahead of cybercriminals.
5. Banks under pressure to protect consumers
Consumers across the world are often the weakest link now, as fraudsters continue to manipulate them in their scams—in what is referred to as “social engineering.” This will drive an ongoing rise in Authorized Push Payments fraud in the UK and other countries. Fraudsters will look to exploit weaknesses in peer-to-peer payment platforms and immediate payments services in 2020—and beyond.
There will also be an increase in scam victims’ conversion to “mules” as customers realize they have been duped and banks refuse to reimburse losses. This will leave fraudsters with a list of victims ripe for further social engineering as they recruit fraud mules willing to have their account used to transfer stolen funds in the hope of clawing back some of the lost funds.
In the same vein, banks will face increasing pressure to protect vulnerable customers from such attacks by identifying threats and proactively defending against them. This is a highly specialized task, requiring accurate demographic profiling and assessment of customers’ financial information in an ongoing manner.
Across all five of these predictions, it can clearly be seen that fraudsters are constantly changing their angles of attack; but financial institutions and payment networks are also finding newer and more efficient ways of dealing with fraud. I look forward to seeing how the landscape will continue to change in the decade ahead, and the new challenges that will develop with these changes.