How does the SWIFT payment network use AI?

Acting on the sanctions, the European countries and the US have unanimously imposed a ban on leading Russian banks from the international payment network SWIFT following the invasion of Russia on Ukraine. The ban means Russia would be in a major economic disruption as it cannot access the global financial market. This means business establishments and firms in Russia will be prohibited from importing or exporting or making any monetary exchange; there will also be restrictions in making any investments or borrowing outside Russia.

So, what is SWIFT?

Society for Worldwide Interbank Financial Telecommunication (SWIFT) is a safe and secure international messaging network that banks from many countries use to make payments from one country to another. This global network connects banks to communicate messages about securely transferring funds using a code. SWIFT was founded in 1973 by a Belgium-based society to facilitate financial and banking institutions to wire money to each other so that global trade could carry on easily and smoothly. SWIFT was set up out of fear that a single private and fully American entity Citibank, formerly First National City Bank, might control the global financial flows.

11,000 banking institutions use SWIFT

About 11,000 members (banking institutions) spread across 200 countries use SWIFT to send nearly 33.6 million transactions every day. In 2020, about 40 million messages with instructions to send trillions of dollars were sent each day over the platform. The banks find SWIFT transactions very safe as the messages are delivered over the network; these messages also help the banking system to honour the payment instructions swiftly. 

Three data centres

Three data centres run the secure messaging network of SWIFT, located in the US, the Netherlands and Switzerland; a command and control centre is located in Hong Kong. These data centres share information in near real-time; however, if there are any lapses reported from any of the data centres, another centre can handle the transaction traffic. SWIFT uses submarine communications cables to transmit its data.

SWIFT currently provides services to financial institutions, including banks, exchanges, depositories, brokerages, asset management organisations, currency exchanges, security dealers and corporate treasuries.

The technology used by SWIFT 

According to Crunchbase, SWIFT uses about 19 technology products and services for their smooth transactions, including HTML5, Google Analytics, and jQuery. About 61 technologies are used for SWIFT’s website, including Viewport Meta, IPhone/Mobile Compatible, and SPF.

How is SWIFT using AI?

SWIFT offers its community a platform for messaging, standards for communicating, and value-added products and services for which AI technologies are used to enhance the quality.

AI for Reference Data is used to 

– Re-structure data files that come from various sources worldwide

– Improve local language translations

– Clean data and provide missing data elements

For example, today, there is a lack of quality in the address field for some countries. SWIFT uses ML to structure that data better and improve the translation for enhanced data quality.

AI for Standards is used to 

– Facilitate the migration to ISO 20022

– Investigate the logic and the pattern among different standards

– Validate new standards

For example, SWIFT performs AI research on different financial messaging standards to facilitate and automate the migration from any financial messaging standards to ISO 20022.

AI for Financial Crime compliance is used to 

– Tune and improve fraud detection rules

– Identify uncharacteristic business relationships

– Optimise effectiveness and reduce false-positive cases

For example, SWIFT performs AI research for payment control solutions to derive a personalised payment rule for each customer automatically.

AI for Business Intelligence is used to 

– Improve data quality

– Enhance products

– Perform profiling and peer group comparison for anomaly detection

For example, for SWIFT GPI, SWIFT performs AI research to determine the quickest, cheapest and most effective route that payment can take.

Although SWIFT has leveraged AI at various stages of their transactions, from messaging to fraud detection rules and detecting anomalous activities in financial SWIFT transactions using ML and behavioural analytics, it is only recently that SWIFT has announced a new AI platform. SWIFT is now leveraging a pivotal role in the financial industry to develop transformative AI solutions by enabling a high-performance AI platform that is future-ready for hybrid cloud.

Transformational AI

At Sibos 2021, Tom Zschach, chief innovation officer at SWIFT, said, “We see that the use of AI and predictive analytics is the key to delivering our strategy, which includes instant and frictionless payments. We are now focusing on building on AI and ML solutions to enhance our existing products. We are rolling out an enterprise-scale AI platform at SWIFT along with our technology partners. Red Hat will help us with the container platform. Kove, a software company, has given us software-defined memory, and C3.AI has given data ops, for machine learning ops and DevOps.”

In December 2021, at the AI Summit in New York, Red Hat and SWIFT presented details about using AI/ML and containers in the financial services industry. As the financial industry’s neutral and trusted provider, SWIFT is helping the community of over 11,000 financial institutions move value around the world reliably and securely at scale.