Visa's AI averts fraud worth $273m in New Zealand banks

Visa, a global expert in digital payments, has announced that through the application of AI technologies, it has assisted New Zealand financial institutions in averting $273 million in fraudulent activity within one year.

The recent figure, which represents a three-fold increase from 2021, emphasises the importance of AI in financial security as fraud tactics become more intricate and advanced, Visa comments.

The disclosure comes in the wake of the escalating rise of phishing schemes around the globe, which have coincided with the advent of Generative AI and other emerging technologies. The first half of 2023 witnessed a marked 40% surge in enumeration attacks, a criminal method that utilises automation to test and guess payment credentials during online transactions.

Visa's drive to use AI technology in risk and fraud management has led to the application of neural networks, modelled on human brain functionality. Visa now has several hundred AI models in operation, powering more than 100 products and services. Amongst these is Visa Advanced Authorisation. This high-speed security system can evaluate more than 500 risk attributes approximately in a millisecond, providing a real-time risk score for all transactions across VisaNet, according to the company.

Joe Cunningham, Head of Risk for Visa Asia Pacific, stressed both the accomplishments and challenges in this area of work. "Preventing more than $270 million in fraud is an incredible testament to Visa's AI-powered security abilities, yet is also a stark reminder that security has to keep pace with innovation," he said.

Cunningham outlined the cat-and-mouse nature of cyber-security in relation to modern fraudsters, stating "As new ways to buy, sell and move money become popular with consumers and businesses, they become equally popular with fraudsters seeking to exploit any vulnerabilities in the system."

Visa also announced plans to present its real-time payment fraud prevention service in New Zealand this year. The aim is to offer further support to financial institutions by providing up-to-the-moment transaction analysis and fast risk assessments. It is hoped this move will continue to help dampen the impact of cyber-related fraudulent activities on businesses, according to a statement from the company.

The real-time payment fraud prevention system, constructed specifically for instantaneous payments, provides a multi-financial institution risk analysis feed. Deep learning AI models are continuously refined to aid financial institutions in detecting potential scam patterns as they unfold. APIs can also receive real-time transaction scores and incorporate this data into existing fraud management tools and processes, reinforcing the existing cybersecurity landscape.