The payment fraud market faces “a number of challenges which makes it uniquely suited to the wide-scale deployment of fraud prevention systems,” according to an update from Juniper Research.
The update from Juniper Research notes that one major feature that AI systems possess is “the ability to scale and deal with exponentially increasing complexity.”
As AI is designed around mimicking the human ability to interpret data by learning from experience, “it is fundamentally more capable than any rules-based system.” The report from Juniper Research added that rules-based systems “invariably lead to high numbers of false positives, as they cannot anticipate every combination of circumstances, whereas AI can adapt to changing conditions.”
The report also pointed out that scale is “the critical element.” While humans and manual reviews are limited by available time, AI capabilities “are only limited by the computational power assigned to the task.”
The report added:
“With the advent of cloud computing, this computational power is barely restricted. As such, AI-empowered fraud detection is capable of far greater scale than traditional rules-based systems, without the large manual work requirements.”
As explained in the report, speed is also a critical factor, for two main reasons:
Time to approve transactions:
In a fast-moving market like eCommerce, which has become “highly competitive, ensuring that merchants are strong on the user experience they provide is more critical than ever.” Given the importance of this, any delay “in which it is unclear whether a payment is going to be approved or not is unhelpful.”
Speed, in terms of “making a decision or requiring additional steps to be taken, is critical to delivering this strong user experience.” AI enables businesses to “cut down decision time when a review is required, and reduces the number of transactions needing reviews at all, reducing the problem further.”
Rapidity of new payment types:
With the rise of payment types that are faster than ever, “particularly instant payments, fraud systems need to be able to intervene and assess transactions in real-time.” The only way to achieve this “is with further automation, which only AI can provide.”
As noted in the update from Juniper Research:
“Pattern recognition is an important skill within AI & machine learning. We define this as the automated recognition of patterns and regularities in data. This is central to how AI functions – by recognizing patterns, as a human analysis can, AI can detect shifts within data and identify irregularities. This capability is essentially what enables AI to deliver within fraud. This capability is important because financial fraud attempts are not isolated – typically, fraud is linked to other accounts and transaction attempts, and has indicators, such as a changed geolocation, a different device or different user behavior.”
For example, within money laundering attempts, “there are often many different mule accounts (where intermediaries are used to pass fraudulent funds) used in a chain, so taking a wider view is important to catching this kind of fraud.”
Fraud can also “come in waves, with new trends emerging.”
As such, “having systems that can identify this and act upon it is important.” Machine learning models “must have the flexibility to change when shifts such as these are detected.”
The global business spend on AI-enabled financial fraud detection and prevention platforms will “exceed $10 billion globally in 2027; rising from just over $6.5 billion in 2022.”
The report added:
“Growing at 57% over the period, we predict that as fraudsters become more sophisticated in their attacks, merchants and issuers will become more adept at utilizing highly advanced AI-enabled fraud detection methods to combat crime. The ability of AI to recognise fraudulent payment trends at scale is critical to provide improved fraud prevention.”
The report also noted that cost savings from AI deployment “will be critical to taking system use beyond regulatory compliance.”
And providing “a genuine return on investment on fraud prevention services, with improving models and greater data access creating a virtuous circle of improvement.”
The team at Juniper says that they “forecast growth of 285%, with cost savings reaching $10.4 billion globally in 2027, from $2.7 billion in 2022.”