Visa's AI tools stopped $354m of fraud in 12 months

Visa’s AI-driven security protocols helped financial institutions prevent more than $354 million in fraud from impacting Australian businesses in the 12 months ending in April, according to estimates from the company.

Visa is using neural networks modelled on the human brain to power its AI technology, which is designed to analyse the risk of transactions in real time to find and stop fraud.

The algorithm is capable of assessing more than 500 risk attributes in about a millisecond to produce a score of the predicted fraud probability of every transaction.

Visa Head of Risk for ANZ and South Pacific Carolina Gallegos said these fraud detection capabilities are proving vital in light of the boom in online retail spending. Data from AusPayNet indicates that online retail spending increased 44% in 2020, but card fraud grew just 0.3% over the same period.

“We have a global team of around 850 Visa employees dedicated to anticipating and countering new threats as fraudsters find new ways to attack,” she said.

“Trust underpins everything we do at Visa, including our approach to innovation. We are investing more heavily than ever in systems resilience, fraud management and cybersecurity, including tokenisation, AI and blockchain-based solutions to bring even more security to e-commerce and Australian businesses.”

The AI technology is being used to counter one of the top payment threats to emerge in Australia and globally the past year – enumeration, a crime that involves using automation to test and guess payment credentials such as account numbers, CVVs and credit card expiry dates during online checkout.

The system has been trained to spot patterns in data that are otherwise undetectable by humans to alert merchants and financial institutions before these fraudulent transactions begin.

“Put simply, commerce is moving online, so fraudsters are focusing their efforts online and Visa’s AI security helps make every single split-second assessment more sophisticated,” Gallegos said.

“A lot of the time, breaches can happen months before actual fraud takes place, so we help businesses identify if they have been a victim early on and take action before it occurs.”

Image credit: ©stock.adobe.com/au/anyaberkut