As fraud attempts and cyber-security incidents continue to rise, more and better Artificial Intelligence (AI) solutions are being touted as the solution. But does AI deliver on its promise?
Two years ago, CEOs speaking off the record at a McKinsey global leaders summit were heard to say they were “done” with AI given significant investments that failed to improve fraud detection rates.
“While successful fraud attacks have decreased, the number of attempts keeps rising and overall dollars lost continue to grow.”
To judge from new research, the fraud problem seems to be getting worse.
A recent study from Aite-Novarica predicted that synthetic identity fraud (fake customer IDs cobbled together from stolen records and used to defraud consumers and businesses) will jump from $1.8 billion in 2021 to $2.42 billion by 2023.
Aite/Novarica also surveyed a group of top fraud executives who pegged “synthetic identities resulting from app fraud” as one of their most worrisome threats.
In 2020, financial institutions suffered $20 billion in losses due to synthetic identity fraud.
The use cases keep piling up: suspicious auto loan applications rose by more than 250 percent, while Buy Now, Pay Later fraud (BNPL) fraud rose by two-thirds over the last year (66 percent increase between 2020 and 2021).
Younger generation most affected
Millennials are more likely than any other generation to have experienced credit card fraud, according to a new study by Morning Consult and IBM.
This may be because millennials use their cards more online, or in high-risk merchant locations, or a host of other reasons.
Whatever the case, there’s no doubt that the number of fraud attempts continues to rise and that the total dollar volume of fraud is also rising – even if the proportion of successful fraud attacks has actually decreased, according to FICO’s European Fraud Monitor.
AI – is it the only answer?
The most recent developments in AI claim immense improvements in fraud identification and prevention are possible.
Using so-called “neural network” algorithms, proponents of AI-based systems say it’s possible to process data at extreme speed (less than 20 miliseconds) across enormous datasets and save up to $5 million per day in potential revenue lost to fraud.
While AI systems claim improvement, new developments in authentication technologies may provide at least half of the answer.
This summer, the FIDO Alliance will be rolling out a version of its authentication system that sits on top of a mobile device or laptop’s systems and does not require a “dongle” or external hardware to operate, while the EMV consortium is soon to launch the EMV 2.3 authentication standard for card transactions.
And these developments are taking place before the EU finalises PSD3, which will itself introduce new authentication and security standards.
From a buyer’s perspective, it would be interesting to run a cost/benefit analysis related to investments made in authentication technologies and those made in AI systems designed to identify and prevent fraud.
While it’s clear these two elements are the “fork and knife” of fraud prevention, more work should be done to determine where organisations are getting best value out of both elements – and what can be done to improve value (as distinct from claimed or real performance) for the companies investing in fraud protection for their customers’ benefit.
The post Fraud and cyber-threats keep rising – is AI the answer? appeared first on Payments Cards & Mobile.