AI joining the anti-fraud beat as financial-services CIOs fight rising fraud
- 01 November, 2017 15:35
Steadily growing rates of fraud are driving APAC financial-services institutions to embrace new technologies at a world-leading pace as they tap into machine-learning tools designed to pinpoint telltale signs of fraudulent activity.
Nearly 1 percent of all verified financial transactions carried across the network of digital-ID verification company Jumio this year were fraudulent, the company announced this week as it debuted a machine language (ML)-based trusted-identity-as-a-service offering designed to curb the behaviour.
Noting that 19.7 percent of fraudulent transactions were due to fake or doctored photos and 19 percent were from altered ID numbers, Jumio – which verifies nearly 200,000 IDs daily – has reworked its Netverify suite to incorporate computer vision-based machine learning algorithms designed to improve its fraud detection.
IBM has also been applying machine learning to fraud detection, this week launching a tool for detecting new account fraud – often used by fraudsters that set up temporary accounts in identity-theft victims’ names, then use them to funnel ill-gotten money. The company’s Trusteer New Account Fraud offering uses ML to help banks correlate a broad range of fraud indicators against the details provided by customers opening new accounts, triggering reviews if factors correlate too closely with known fraud indicators.
With Attorney-General figures suggesting that identity crime is costing Australians more than $2.2b each year – including $600m lost to credit-card fraud, identity theft and scams – better fraud detection is proving critical in limiting the exposure of companies that are increasingly moving core administrative transactions online.
The latest figures from the Australian Cybercrime Online Reporting Network (ACORN) showed steadily increasing incidence of fraud in this country, with 6102 cases of scams or fraud reported during the quarter to September. This was up 12 percent from 5432 cases during the same period a year earlier, and up 42 percent from the 4296 cases reported two years ago.
Offering ID verification as a service will improve the ability for other services to incorporate better fraud detection using straightforward API calls – of particular interest for banking and financial-services companies that have been rushing to adopt ML and artificial intelligence techniques.
AI has quickly grown to primacy as a way of improving fraud detection and supporting transactions that are increasingly being conducted completely online, often from mobile devices.
A recent National Business Research Institute survey, for example, found that 62 percent of financial-services organisations were planning to adopt AI into their business processes by 2018. And some 38 percent of APAC respondents to Teradata’s recently-released State of AI for Enterprises study said they expect AI to have a strongly positive impact on the financial-services market – putting it second only to IT, technology and telecommunications.
Fully half of APAC respondents saw AI as valuable in finance transformation – transformation in financial auditing, regulatory compliance and reporting – compared with 41 percent globally, 40 percent in the Americas and 37 percent in Europe.
Strong APAC support for AI validates recent observations that CIOs inside of APAC companies are adopting disruptive technologies faster than their peers in other areas.
Gartner’s 2018 Gartner CIO Agenda Survey, which included 3160 respondents, found that 37 percent of APAC CIOs had deployed, or are about to deploy, AI solutions – well ahead of the 25 percent global average. IT budgets in the ANZ region are expected to grow by 3.2 percent on the back of new investment, up from 2 percent last year.
There are signs that the surge in investment is paying off: Verizon’s Data Breach Investigations Report 2017, for one, noted an increase in fraud detection in industries like accommodation and food services – where fraud detection was responsible for detection of 85 percent of all breaches, typically through point-of-sale (POS) systems. That figure was 25 percent higher than the previous year, reflecting the increasingly effective anti-fraud measures now being put into place.
Delegating fraud detection to external third parties offers considerable value for transactional companies because it offers the potential to detect and block transactions before they are completed, Verizon noted, “whereas internal audits discovered the fraud after the proverbial horse had left the stable”.