Long settlement times. Intensive compliance requirements. The threat of fraud. Despite a pervasive shift to digital transactions, cross-border payments providers and their customers are still up against a complex slate of issues.
So how can the industry tackle these challenges and mitigate delays for businesses that need to efficiently send funds around the world?
Enter: artificial intelligence (AI). Whether automation is framed as a workforce revolution or an existential threat, the last few months alone have made it clear that the technology is here to stay. And where FX payments are concerned, there are already several clear use cases — meaning AI implementation may soon be a major advantage for traditional financial institutions and modern fintechs alike.
Detecting fraud in the era of real-time payments
The advent of instant payments is ushering in a new level of convenience for both B2B and personal transactions at a global scale. However, this innovation adds a new layer of difficulty to preventing fraud.
What makes it so hard to ensure real-time payments are carried out securely? According to analysts from J.P. Morgan, cross-border payments are especially vulnerable to cybercrime, with fraudsters exploiting opaque regulations and inconsistent messaging standards across regions. There is also widespread targeting of accounts payable operations through the use of fake invoices or illegitimate supplier accounts.
These factors illustrate how the speed that businesses and customers have come to expect through near-instant payments doesn’t always square with the realities of fraud mitigation.
“It’s much harder to reverse payments than it is to verify them,” said Scott Johnson, VP, Program Management at Convera. “Payments providers need to be incredibly thorough upfront.”
One primary tactic to detect fraud in FX payments is pattern recognition, such as comparing a transfer request to customers’ past behaviors, schedules, and common recipients. Sifting through this data, though, is often incredibly labor-intensive and time-consuming — posing a serious obstacle to clearing instant payments.
But with AI and machine learning (ML) technologies, payments companies have the power to automate data reconciliation, in turn flagging suspicious FX payments before it’s too late. This also allows specialists to devote time and resources to requests that raise concerns, rather than each transaction.
A new compliance tool on your side
Compliance is another area that AI promises to simplify, speed up, and strengthen.
Traditionally, compliance and sanctions proved especially difficult for FX payments providers. As Grant Vickers describes in the Payments Journal, cross-border payments involve “bridging multiple currency systems and regulatory jurisdictions, and generate far more sanctions alerts.”
When Russia invaded Ukraine in early 2022, for example, sanctions screening became much more complicated overnight. The US imposed sanctions that affected roughly 80 percent of banking assets in Russia, while SWIFT also restricted access to its network.
With ongoing geopolitical tensions and ever-changing regulations, maintaining compliance will remain a challenge. By integrating AI, however, the FX industry won’t need to be stuck playing catch-up.
As Johnson of Convera outlined, certain inputs — such as someone’s birthday or address — can quickly show that an individual is on a sanctions list. Yet there will still be many exceptions or complicated cases where analysts need to make a final decision. The more data-driven recommendations they have at their disposal, the more accurate and efficient that process will be.
“Leveraging AI has the potential to create a safer fabric of payments for everyone and empower people to make better decisions,” added Johnson.
What comes next for AI in FX payments
The applications of AI are still evolving, but concrete developments are set to make its capabilities in analyzing data and recognizing patterns even more valuable. Among the most promising updates is the increasingly widespread adoption of ISO 20022, a messaging standard for exchanging electronic messages attached to payments.
ISO 20022 makes it easier to provide rich data with every transaction and offers more detailed information about each stage of clearance among intermediaries in the payment life cycle. Though migration to the system among financial institutions is ongoing, estimates from SWIFT show that 79% of the total payments worldwide volume will use the standard by the end of 2023.
Most importantly, a common messaging format may bolster AI systems and drastically reduce false alarms for fraud and sanctions. Achieving higher accuracy at scale would be a significant advancement — especially given that, on average, each false sanction alert takes between three and five minutes to resolve manually.
“AI could really be a self-service solution for cross-border payments companies in the long run, giving analysts relevant insights and more useful recommendations,” said Johnson.
While there’s no substitute for human expertise in the FX industry, it will take a new toolkit and a wide range of technologies to meet the demands of global finance today.
Small firms embrace AI – a growing trend