Sibos 2018: Need for speed – ML reducing fraud in real-time payments
Machine learning (ML) must have broad application across the growing network of real-time payment pathways to reduce the opportunity for fraud, according to experts at the day three Sibos session on combatting fraud on real time payments.
“ML or decision science is key. The only way to beat real-time fraud is to play in real time and this can only be done on an ML basis. We need to be predictive and not reactive,” said panellist David Rich, executive vice-president, Mastercard.
“The only way to beat real-time fraud is to play in real time”
Fellow panellist HSBC’s global head of correspondent banking, Barbara Patow, explained the bank uses ML to screen 1.2 million payments a day and stops 75,000 payments. Of these, 35,000 are closed, although staff are involved in these decisions.
Panellists stressed the need to be able to detect fraud before it happens in the real-time payment environment, and voiced concerns about a potential rise in fraud as a result of instant payments.
Rich told the audience there is already $2 trillion in illicit funds in the global financial services system and real-time payments will exacerbate this problem.
“Criminals are very organised and they collaborate, constantly probing for the weakest link,” he said, explaining how Mastercard is often able to trace frozen funds involved in money laundering back to the source and repatriate it.
Australia’s New Payments Platform (NPP) has been designed to facilitate real-time payments and as such, banks are introducing new ways to reduce payment fraud.
“There’s a real focus on fraud in the NPP world,” said Westpac’s global transaction services general manager, Di Challenor. She explained how the bank is able to recognise where a device such as a smart phone is when a customer opens internet banking. If the phone is used in a way it’s not normally used, the bank receives an alert to indicate a fraud may be being perpetrated.
Although Challenor acknowledged consumers may view this as intrusive, they look to banks for fraud detection.
“The NPP has rich data and great benefits, but as a bank we have to assess it and ensure there is no financial crime risk. We’re looking at digital IDs to protect money movement and citizen information,” she told the audience.
Panellists called for an industry-wide response to tackle the issue of fraud in real-time payments.
“We need to bring regulators on the journey and as a community work together and evolve together,” said Patow.
Additionally, NPP Australia’s general counsel and company secretary, Vanessa Chapman, said ensuring the platform and market participants comply with minimum standards will help give counterparties in a transaction assurance they won’t be unduly exposed to fraud.
“Good prudential supervision and reasonable standards allows you to trust each other as banks,” she said.
By Daily News at Sibos contributor Alexandra Cain
This article is also featured in Daily News at Sibos 2018, our flagship daily publication at the Sibos conference.
Click here to see the Daily News at Sibos editions – they are all free to read!
Follow us on Twitter @DailyNewsSibos
