By Jimmie Franklin | Published in Vixio
Artificial intelligence (Al) has never strayed far from the headlines in 2023, but how is the payments and banking industry using it, and how are regulators responding?
For every positive sentiment about Al, there has also been a negative.
Yet concerns about the fast-growing technology have not stunted intrigue and investment from the financial services industry, including in payments.
“Al is now being used just about everywhere, and momentum has cranked up to 11,” said Oliver Tearle, head of innovation technology at The ai Corporation. “There is huge amounts of activity in the fraud space, but also in automation, advertising, and analytics.”
Tearle noted that there will be other areas that gain traction in the future.
“In spite of concerns about the speed of development recently, I don’t think that Al development will slow down,” added Tom Whittaker, associate at Burges Salmon law firm.
Companies see effective use of Al as a competitive advantage and they will look to have risk management in place for this, he said. “The thing that is most on companies’ minds at the moment is how they can best leverage Al for business purposes.”
Whittaker continued that nobody wants to be left behind. “There has been so much talk in the market about what Al can do, and we’re seeing companies integrate large language models and develop their own models.”
“Al is being used all over finance currently, from financial planning and analysis to risk products, and much more,” pointed out Maya Mikhailov, founder at San Francisco-based SAWI Al.
“On the consumer side there are many places where Al can be used to enhance customer experience in personalisation, for example, suggesting offers a consumer might be interested in or a term loan that can best meet their cash flow needs,” she noted.
However, Mikhailov noted that Al also has a vital role in streamlining operations and making companies more efficient. “For example, it can make the budgeting and revenue prediction process of a financial planning and analysis department faster and more accurate by using constantly evolving models.”
In payments, thus far, Al use has gained traction for one of the industry’s bugbears – financial crime prevention.
“Al is being used all of the time in the fight against fraud,” said Tearle.
This includes a lot of different models for fraud automation, which goes all the way from basic fraud screening at the merchant to the issuing bank, he noted.
“This happens before authorisation happens, as well as after, to verify and to prevent future cases of fraud.”
According to Craig Walker, chief product officer at Versa pay: “There is a high cost to fraud that impacts not just financial institutions but also the customers that they serve.
“Using Al to reduce fraud, limit risk, and automate processes ultimately leads to a reduction in costs associated with both,” he suggested.
For example, Walker noted that Al may start to put pressure on card interchange rates as fraud rates drop due to the automated intervention of Al, driving down the cost of acceptance.
And what do the regulators think?
In this fast-moving world, where payments and banking players are increasingly investing in new technology, regulators too have needed to keep on top of market developments.
Some progress has already been made among some forward-thinking regulators. The EU, for example, is in the process of negotiating its Al regulation.
Meanwhile, the UK government is due to hold a global Al summit in September.
“We’re increasingly seeing the rise of Artificial Intelligence (Al) and the interlock between data and payments, Blockchain, open banking, and a range of other instances of technology and payments morphing,” a spokesperson for the UK Payment Systems Regulator (PSR) told VIXIO.
The PSR spokesperson said that, as the world evolves, it will continue to monitor the wider payments horizon to understand developments and their potential impacts.
“We want to encourage competition and innovation, while making sure payment systems offer access, protection and choice for everyone that uses them.”
Regulators are also now using Al like many of the entities they supervise to increase efficiency.
A spokesperson for Germany’s Federal Financial Supervisory Authority (BaFin) told VIXIO: “In the ongoing supervisory practice, Ba Fin is already using Al applications and/or algorithmic models to some extent, for example in security supervision, when it comes to transaction monitoring and automated identification of insider trading as well as market manipulation.”
The regulator also uses automated pre-evaluation tools for the analysis of textual reports.
The spokesperson said that BaFin is continuously exploring new technologies and possible areas of Al application in its supervisory practice.
“A dedicated team of experts has been established to foster data and analytics-driven supervision.”
This team supports and ensures the technical and organisational framework within Ba Fin to enable the business unit to further develop database supervisory instruments.
As for those that it supervises, BaFin said that it is tech-neutral, and thus Al and machine learning (ML) procedures are regulated in the same way as traditional methods.
“From our point of view, existing regulation already covers these algorithms to a large extent.
“Many of the aspects associated with financial institutions’ use of ML/Al are quite similar to those posed by traditional models, especially considering that a large part of ML/Al applications in finance are not technically new but have existed in the form of statistical modelling techniques for a long time.”
However, the regulator did acknowledge that supervisory practice, such as the methodology for on-site investigations, needs to adapt to these new developments.
“In any case, firms’ risk management and governance processes are expected to identify, monitor and address any specific risk arising from ML/Al and have to ensure a high level of consumer protection.
“While applying ML/Al tools supervised institutions are expected to ensure that they are used in a responsible way.”
Ba Fin appears to be ahead of other regulators in this regard.
A Bank of Lithuania spokesperson told VIXIO that they use non-Al tools for their supervision currently, but this may be about to change.
“The increasing data flows and the prospective demands require us to look for appropriate solutions,” the spokesperson said.
One of these might be a regtech solution for data collection, which is currently being developed by the Bank of Lithuania in the context of ITS Data Management Maturity (DAMAMA) programme development.
“This solution would collect AML or other data for supervisory purposes from electronic money or payment institutions directly, analyse it and provide data,” the spokesperson said. “It would return aggregated data with the analysis which would be open to market participants.”
The solution itself was presented in 2020, and now it is being tested for production.
“As a relatively new and very dynamic technology, Al provides great opportunities for entities and consumers,” the Bank of Lithuania spokesperson said.
However, they added that the regulator is aware that this could result in undesired outcomes if risks are not properly managed.
“For instance, these risks could include privacy breaches, disclosure of sensitive information, manipulation of customers, and biased outputs,” the regulator told VIXIO.
With regulation veering closer worldwide and companies increasingly seeing new opportunities, it seems likely that Al will be a top priority for regulators.
The big concern from firms is whether regulatory proposals will negatively impact the methods that Al is currently being used for to improve business processes, products and services.
View the article in Vixio here
Vixio is the leading provider regulatory intelligence to the payments industry. Visit their website here