Fortiro automates fraud detection processes, cutting processing times while enhancing overall information security. Learn more.
Reports in the second quarter of 2021 showed a 37.2% year-over-year increase in mortgage fraud risk, with an estimated 0.83% of applications – roughly 1 in 120 – containing fraudulent documentation. For financial institutions in the market for fraud detection solutions, the Fortiro Protect platform – founded by PwC Australia in 2019 – has emerged as a frontrunner in applications of state-of-the-art technologies to fraud detection and identity verification in loan applications.
Fortiro applied technologies previously unavailable in the industry to bring anti-fraud measures into the 21st century, helping financial institutions verify in real time if payslips, bank statements, or other documents are valid or fraudulent. These technologies include:
Traditional fraud detection methods rely on fraud teams manually reading and inspecting all the applicant’s documents for common known indicators such as spelling mistakes, unusual fonts, irregular formats, and discrepancies between numbers reported in different documents. While employees can develop some skills in these areas through experience, the process remains time-consuming and always involves a significant number of hunches and guesswork.
Fortiro automates most fraud detection processes, cutting turnarounds down to a fraction of typical times. Using OCR for data extraction, the platform analyzes documents along three axes: text, images, and the location and structure of text. Fortiro’s proprietary AI then scans documents for layout anomalies and applies labels to data points.
With data points identified, the platform performs three machine learning-enabled fraud detection functions:
Beyond these automated functions – which Fortiro completes in a matter of seconds – the platform integrates its internal analysis with a suite of third-party data sources for income verification such as ABN lookups. Finally, Fortiro adds automated name matching, policy checks, and derived applicant income to the file before assessing whether to apply a high-risk flag and initiate an investigation. Users of Fortiro can also configure custom flagging rules based on their organization’s needs.
Other time-consuming manual processes in loan applications include those performed to verify income and redact sensitive information where necessary. Typically, this process requires staff to collate and compare different applicant documents to verify that reported incomes and assets match in different places and that the client’s proven income meets overall creditworthiness requirements. If clients have complex transaction histories or multiple irregular sources of income, delays and manual work hours can accumulate rapidly.
For client income documents such as payslips, Fortiro’s proprietary OCR-enabled algorithm will perform income extraction matching and transaction extraction analytics. Many other document data extraction services offer OCR-generated readouts of applicant documents. However, what Fortiro brings to the table as an industry-first solution is machine learning processes that enable high-accuracy labeling for critical data points.
To explain, payslips are unstandardized, and any technologies applied to interpreting them automatically must be capable of identifying unique and previously unknown formats. Any confounding numbers read in unexpected sequences, such as pay dates, net pay, or gross pay, will result in meaningless extractions. Fortiro excels in these tasks compared to other OCR-based technologies and can reliably produce complex extractions such as correctly identifying payslips in bundled PDFs containing dozens of client documents in a single file.
Businesses across industries experienced heightened fraud attempts and cybercrime during the COVID-19 pandemic as organizations rushed to adapt to rapidly implemented remote work policies. As more sensitive data was brought online for the first time in third-party cloud services and face-to-face interactions were suspended in many business contexts, opportunities for cybercrime and fraud proliferated by as much as 600% globally.
Financial institutions in Australia and New Zealand have been no exception in this regard. Among surveyed institutions, 65% reported increased fraudulent activity linked to remote work vulnerabilities and 25% experienced delays or cancellations in their existing fraud and corruption prevention programmes.
While pandemic-related restrictions have largely expired, remote work in a significant portion of the workforce appears to be here to stay and will likely increase in 2023. For financial institutions, this means that the burden of preventing fraud in liar loans and other cybercrimes will only become more challenging. Getting ahead of this curve means adopting new technologies and adjusting to new information security practices.
Fortiro Protect can play an integral role in this necessary adaptation. Combined with other cutting-edge technologies, Fortiro’s automated fraud detection capabilities can help organizations protect themselves in an evolving threat landscape.
Fortiro has recently joined forces with the secure file sharing and document collection platform FileInvite, enabling clients to target the two major processing bottlenecks – document collection and fraud prevention – simultaneously. With combined bank-grade security controls to protect client data, Fortiro and FileInvite drastically reduce your organization’s processing times while enhancing overall information security and preventing fraud.
To learn more and request a demo, visit FileInvite today.
With CDD requirements becoming more complex, financial institutions must understand the different kinds of CDD processes. Here are 5 types of CDD...
How do you receive large files from clients? Here's our 5 step guide to receive large files. Using FileInvite can really help streamline the process.
HR contract management and employee onboarding doesn't need to be complicated, automated systems are here to make these processes much more efficient.