JPMorgan, Commonwealth Bank combine AI tools to improve payments

Brendon Thorne/Bloomberg

The increasing speed of payment processing and the growth of cross-border payments are heightening the risk of mistakes or fraud, a problem Australia's Commonwealth Bank and JPMorgan Chase believe can be countered with a mix of advanced artificial intelligence and distributed ledgers.

The two institutions are connecting products they developed separately to validate bank-account details for international payments. The team up is part of CBA's attempt to extend its payments technology to other banks and JPMorgan's strategy to grow its payments franchise as the bank changes payments leadership.

Like most banks, JPMorgan and CBA are challenged to embrace real-time or faster transaction processing and grow digital digital commerce, while safely and economically moving funds among parties in different countries.

"One of the most important things that consumers do is move money. We want to make sure the user doesn't have a bad day," said Ethan Teas, executive general manager of payments at CBA.

JPMorgan's product, Liink, supports exchanges of peer-to-peer information between organizations such as banks, credit unions or fintechs, with a focus on international transfers. JPMorgan's blockchain unit Onyx designed Liink, which acts as a complement to the Swift messaging service to improve processing for cross-border payments.

"Failed and fraudulent payments are an increasing threat to financial institutions, corporations and customers globally," said Umar Farooq, co-head of global payments sales and CEO of Onyx, in an email from JPMorgan's public relations office. "We developed Onyx and its Liink network to bring the ecosystem together to solve payment-related issues in a collaborative way by first breaking down data silos."

JPMorgan Chase

How JPMorgan is developing an internet of money

July 25, 2022 1:46 PM

CBA's program, NameCheck, uses AI and algorithms to search account details for first-time payments, then matches those details against available payments data to look for possible scams or problematic typos. If there is a mismatch, the user is prompted to take additional steps to ensure the payee is the intended recipient. CBA is looking to pilot the sharing of limited payee account data through CBA's NameCheck technology on the Liink network.

The Australian bank CBA introduced NameCheck in 2023, and the bank claims the product has saved about $300 million within the first year in either fraudulent transactions or payments with inaccurate information.

"It's about how you can get smarter about the context around the payment," Teas said.

For any transaction, but particularly for cross-border instant payments, verifying the account information will become increasingly critical, according to Gareth Lodge, an analyst at Celent.

"[The CBA/JPMorgan partnership] does definitely add value to the Liink network," Lodge said. "It becomes a differentiator for anyone looking to send payments to Australia. Whether the blockchain will make it better starts to then become more of a debate, and you'll have plenty of people on either side."

The use of AI in NameCheck builds on other AI projects at CBA that accumulate payments data and use machine learning to provide more detailed information about financial transactions. The bank's goal is to not only mitigate fraud risk, but also use AI to improve user experience and transaction-processing speed and accuracy.

For example, CBA is using improved language modeling to spot language that is potentially tied to abusive payment behavior, which can often be subtle. CBA is also using generative AI to create virtual consumers to test new products. Gen AI draws on past behavior to produce responses to products that are "informed," and is implemented alongside more traditional product testing

"We're using AI in a number of areas, mostly to identify payments that can move safely," Teas said. "The use of abusive language in payments is a big problem and gen AI can hopefully help the banking market uplift controls to fight that problem."

Other payment companies are also using new forms of AI to combat transaction fraud. In late May, Mastercard updated its gen AI technology to identify account numbers for stolen credit cards. The AI pulls from a growing trove of payments data to construct account numbers that are otherwise shielded from view on the dark web. Visa also recently introduced AI-powered tools to improve risk management.

"Confirmation of Payee, or CoP" has become a necessary tool against scams, especially targeting real-time payments, according to Cleber Martins, head of payments intelligence and risk solutions at ACI Worldwide.

A scam or other type of fraud, like scammy QR codes, can lead the victim to send money to an unintended destination, according to Martins.

"For example: You believe you are sending money to a friend, but the instructions lead to another friend's account. You don't know, unless there is a step prior to your confirmation, that inquires, 'Are you trying to send money to friend No. 2?', giving you the opportunity to confirm or stop," Martins said.

The services that JPMorgan and CBA developed are designed to find this information, so the bank can alert the customer, Martins said. "This information may come from multiple sources, but most common is to connect with the bank that owns the payee account for validation," Martins said.