Anywhere Fraud
Now featuring OrbNet Forensic AI to deliver 95%+ prevention rates on targeted fraud use cases.
Anywhere Fraud
Anywhere Fraud is a specialized module within OrboAnywhere designed to detect check fraud by interrogating the attributes of check images to create a risk score based on the probability of the item being a counterfeit check, forged signature, or altered check. Building upon these scores, the system uses self-learning account profiles to create a statistical representation of the account’s check writing habits and styles.
On-us and deposit fraud has become a focus of the financial industry. 2018 delivered a major spike in both check fraud attempts and losses, as the ABA Deposit Account Fraud Survey from the American Bankers Association estimated 2018 overall check fraud attempts at $15.1B and losses at $1.3 billion…illustrating a major issue with counterfeit checks, forged checks, and altered check fraud.
There is one technology which can combat this trend; integrated image analysis for check fraud detection. The solution: OrbNet Forensic AI!
Anywhere Fraud with Image Forensics can be utilized directly by banks as part of their fraud detection technology set or integrated within a service bureau/partner/vendor's fraud review platform or solution suite for detection for on-us fraud.
Check Fraud Still a Persistent Problem
In 2020, OrboGraph’s survey confirmed that the concerns over check fraud continued, as the major of respondents anticipated fraud attempts and losses to increase in 2021 (See Exploration of Check Fraud in 2020). These risks associated with check payments created a major market opportunity for financial software vendors, payment fraud and risk technology companies, outsourcers and service bureaus.
Several reasons stand out as the driving force behind the dramatic increase in check fraud attempts and losses including:
- The deployment of EMV, which refocused fraudsters on check payments.
- The expansion of mobile and RDC deposits with higher deposit limits.
- Check fraud systems deployed which rely on dated data analytics, without verifying image attributes.
- A focus on real-time and electronic payments with less emphasis on check payments.
- New fraud schemes involving COVID-19 payments: See examples on Modernizing Omnichannel Check Fraud Detection blog.
Integrated Image Analysis
OrbNet Forensic AI is OrboGraph’s new AI and deep learning model approach to boosting fraud detection on check images to over 95% on targeted fraud use cases. Using AI for digital image analysis for the extraction of meaningful information from images, targeted image-based analyzers are able to detect counterfeits, forgeries, and alterations in situations where transactional analysis or machine learning alone fails.
Anywhere Fraud now relies on OrbNet Forensic AI to perform the following tests for the detection of fraud:
- Check Stock Validation (CSV-AI): Analyzes the attributes, layout, and relative coordinates and dimensions of select preprinted fields as “anchor points” on the check. It does not include aspects of the check that are traditionally completed by the check writer or payor/maker.
- Automated Signature Verification (ASV-AI): Identifies signature variations by comparing the signer on the check presented for payment against the signature in cleared checks drawn on the same account. OrbNet Forensic AI activates 512 feature vectors on every signature to analyze attributes as a “forensic screening” process. See Forensic Document Examination interview for more details.
- Writer Verification (WV-AI): Handwriting analysis leveraging OrbNet Forensic AI to identify difference in writing styles.
- Alteration detection using check style comparison and amount discrepancy: Detecting alterations is composed of two sub-analyzers: Check Style Analysis and CAR/LAR Discrepancy.
Transaction Analysis (Highly Targeted Data Analytics)
Transactional Analysis is the process of detecting check fraud by analyzing the serial numbers and amounts on the checks and applying a set of transactional analyzers looking at transaction attribute. This can also be an important component to the system for standalone environments.
Combined with the power of self-learning, account-level profiling, targeted image analyzers are applied after account profiles learn the behavior of the account holder by analyzing both the image, as well as transactional characteristics of the account.
Note that image and transaction analysis can be applied to any workflow as a way to detect fraud prior to posting. For example, inclearing items are the focus of on-us fraud detection, while transit checks are a part of a comprehensive deposit fraud approach.
In dealing with fraud detection, low suspect rates (false positives) blended with high detection rates (false negatives) are the most critical objectives for an optimized system. For a full list of image and transaction analyzers, see Anywhere Fraud on-us brochure.
Additional Features
- Individual analyzer scoring is available for optimal tuning and threshold management. Complements existing rule-based systems as well.
- Match to known fraudulent images which were previously confirmed by a fraud analyst.
- Profiling of multiple R&T on-us numbers and transit R&Ts: prevents a wide range of on-us and deposit fraud use cases
- Real-time interface from any deposit/capture channel across the omnichannel: delivers instantaneous fraud detection across traditional and self service workflows
- ASV on deposit tickets: matches signatures on tickets to customer profiled signatures
Anywhere Fraud is built on the foundation of targeted analyzers performed on an account level. These account profiles leverage self learning algorithms to adapt to the image and transactional characteristics of each account. The system builds a history of image snippets and transactional data of check writers for batch or real-time fraud detection capabilities.
Targeted Deployment
Existing or new check processing business partners, service bureaus, and financial institutions can implement Anywhere Fraud via a batch or real-time web services interface. Additionally, a file import process with “landing zone” approach can be deployed if programming resources are not available.
This flexibility allows Anywhere Fraud to be deployed into any traditional or self-service check processing workflow including Inclearings, teller image capture, branch image capture, ATM, mobile RDC or RDC. So depending on the fraud use cases, the system is optimized around the requirements of the environment.
Note that either an existing suspect review platform or the check capture system is utilized to review suspected fraud items along with companion images delivered by Anywhere Fraud. This consolidated review workflow is very efficient as it does not create additional overhead or system administration.
Business Case Fraud Detection Benefits
- On-us checks (Inclearings, over-the-counter, and self service channels)
- Counterfeit transit checks
- On-us check cashing
- Transit check cashing
- Protect on-us customers from fraudulent events
- Improved interbank check clearing by replicating loss prevention practices
- Reduction of false positives to streamline suspect review process
- Improve overall check fraud detection
