Visa's payment police saves Aussie businesses $560m

Visa said increasing use of artificial intelligence systems to track all of its transactions has enabled the global payments giant to prevent more than $560 million worth of fraud from hitting Australian businesses in the last financial year.

The company has invested in internally-developed AI-based security capabilities since 1993, and boasts that its Advanced Authorisation capability is now one of the most sophisticated non-military applications of the technology in the world.

It uses systems based on neural networks, modelled on the human brain,
to analyse all transactions in real time and can judge if they look fraudulent in a millisecond.

Visa’s group country manager for Australia, New Zealand and South Pacific, Julian Potter, said the $560 million figure was the result of a decline in the number of suspected fraudulent transactions in Australia in the 12 months to June 30. It was about $400 million the year before.

"The main impact of lockdown has been the acceleration of payments up the digital curve. For retail payments, this is highlighted in the shift to online commerce which presents a new opportunity for businesses," he said.

"The amount of fraud prevented has grown since last year, but there’s two sides to this – the amount depends on the level of cyber criminal activity at any moment in time, as well as the effectiveness of prevention.

"With AusPayNet reporting a reduction of fraud rates and Visa’s fraud prevention being where it is, this is good news for Australian merchants."

The data released by payments industry group AusPayNet in August, showed card fraud costs falling in Australia, with locals being stung to the tune of $464 million last year, down from a peak of $576 million in 2018.

Those numbers meant fraud rates had dropped from 75¢ per $1000 of card spend, to 56.6¢ - levels last seen in 2014.

Visa said it processed more than 138 billion transactions in 2019 and had used its AI capabilities to analyse all of them.

Mr Potter said one of the toughest challenges in payments was separating good transactions made by account holders from bad ones attempted by fraudsters, without adding friction to the process or declining sales unnecessarily.

The Advanced Authorisation software works by assessing each payment, based on more than 500 risk attributes, which are known to indicate potential fraud.

This can include whether the account has been used at the store in the past, what type of transaction it is, such as in-app or online, contactless, chip or magnetic stripe, whether the purchase took place at a usual or unusual time of the day, and whether it was for an unusually large amount of money.

"The AI assesses these attributes in about a millisecond to produce a score of the transaction’s predicted fraud probability, with a score of one being the least risky, and a score of 99 being the most risky," Mr Potter said.

"At an individual level, Visa’s AI doesn’t know a consumer’s name or exactly what they’ve bought, but it sees activities and patterns that enable it to understand purchasing behaviour and protect consumers from fraudulent use of their accounts."

The software is part of a suite of AI-based security protections developed by the company, which also includes a Risk Manager that lets financial institutions set rules to decline high-risk purchases that need further investigation to judge if they were made by the real account holder.

It has also just launched Visa Smarter Stand-in Processing, a new capability that uses AI to help financial institutions manage transaction authorisations when service disruptions occur.