Case Study

Using AI to Protect Veterans from Fraud

Mobile Menu

The Challenge
The Approach
The Solution

Keeping fraudsters away from veterans' benefits

The U.S. owes much to its veterans and the extraordinary sacrifices they’ve made. The Veterans Benefits Administration’s (VBA) fraud-prevention work is central to ensuring that incalculable debt is honored to the fullest.

Delivering disability compensation, pension benefits, and other services to millions of American veterans, VBA strives to provide them with timely, excellent customer service, and to make it as easy and frictionless as possible to apply for and receive their benefits. But at the same time, VBA’s Benefits Delivery Protection and Remediation (BDP&R) division must also do its utmost to make the agency discouragingly hostile to criminals who would attempt to divert and steal the funds that veterans and their families deserve and rely on.

As fraudsters and their schemes grow in number and sophistication, VBA is responding by partnering with Booz Allen to implement AI and advance the systems and processes it uses to detect and combat fraudulent activity.

The Challenge: As Fraud Grows More Sophisticated, Anti-Fraud Programs Must Follow Suit

VBA has historically used analytical methods to detect fraudulent activities aimed at veterans’ compensation, pension, and other benefits payments. Unfortunately, over time these methods were becoming less effective than they were in the past. To stay ahead of fraudsters who are embracing more advanced tactics and technologies, VBA needed to respond in kind.

Understanding this, VBA’s leaders partnered with Booz Allen to modernize the agency’s approach to fraud, waste, and abuse prevention.

The Approach: Leveraging AI and Advanced Analytics in the Fight Against Fraud

Working together with VBA’s BDP&R team, we identified several improvements that would increase the efficiency of the agency’s fraud detection capabilities, shore up confidence in the program, and keep veterans’ benefits protected now and into the future. By improving the analytics driving VBA’s various fraud heuristics and applying advanced AI and machine learning techniques, we increased accuracy by reducing false positives, generated manageable reference lists, and helped focus investigation and remediation efforts on value-add activities to ameliorate fraud cases and proactively make veterans whole again without disrupting their lives.

The Solution: Fully-Modernized Fraud Protection Systems for the Nation’s Veterans

Booz Allen and VBA teamed up to develop analytical techniques tailored to the specifics of the various types of fraud that are most relevant to the agency.

For example, to help VBA improve its vigilance against suspicious IP addresses, we developed scripts to match and flag such IPs. We now monitor the resulting list for irregular activities. Additionally, to stop payments from being fraudulently claimed on behalf of deceased veterans, we built a new analytical tool that uses matching logic to identify improper payments with close to 100% accuracy. To help VBA filter out false positives from its suspicious-payment-redirect fraud investigations, we built a machine learning model to accurately identify instances of actual fraud.

These efforts combine to improve VBA’s confidence in the reliability of its fraud detection systems and increase the accuracy and efficiency of the BDP&R division’s work. By strengthening VBA’s ability to detect fraudulent activities with sophisticated analytics, we’re improving the integrity of the benefits process and advancing the agency’s overall mission of bettering the lives and circumstances of the nation’s veterans.