Page updated on October 24, 2024
Economic Services Administration (ESA) – Support enhancement of anomaly detection using AI while maintaining a human-centered focus before taking proactive fraud prevention action.
The Economic Services Administration (ESA) transforms lives by empowering individuals and families to thrive. We serve many clients who live on the margin. Although most are not receiving a cash grant, they may be relying on food assistance, work-related support services, assistance with child support or medical coverage. A family crisis or change in the economy, even a small one, can force these families into situations requiring assistance, whether it’s the full support of a cash grant, help with child support or temporary assistance to avoid losing housing.
Vision
Increase capacity in the state's ability to utilize AI while maintaining a human-centered focus in a proactive fraud prevention approach. This approach will aid in the early identification of improper benefits application submittals and usage of public assistance benefits.
Key Challenge
Preventing inaccurate eligibility benefits for clients. Timely detection and effective prevention of fraudulent benefit issuance and use.
Existing Fraud Prevention Products & Processes
- Areas of fraud identification and early prevention include criminal actors’ use of stolen or fabricated identities, stolen EBT card information, terminal transactional anomalies, benefit fraud, and insider threat.
- Washington State has one EBT vendor (FIS) and the state utilizes data sources available from the vendor to support its proactive fraud prevention efforts.
- The agency's fraud prevention unit has criminal actor fact patterns identified through the use of data analytics they are currently following. These fact patterns continue to evolve and change over time. The relative fraud rate remains low in perspective to public benefit caseload data.
- Historical data consists of fraudulent fact patterns that are utilized in proactive fraud prevention efforts.
Primary Customer/Users
- DSHS employees – the current team consists of approximately four individuals who focus on proactive fraud prevention efforts, but this can expand based on the needs of the awarded solution.
- CSD, DPI and OFA staff involved in Fraud Referral process (CSD/DPI) and fraud investigation and response (OFA).
- Customers seeking benefits
Security Requirements
- A Security Design Review will be initiated before acceptance
Required Technology (Compatibility)
- Real-time connection with multiple internal and external data sources
- Ability to quickly process millions of lines of data in a real-time environment to quickly identify anomalies
- System competency assessment for accuracy, efficacy and EDI-compliance outcomes. Details on System Compliance will be shared with the successful vendor. Review this page for more information.
- Maintenance & Operations (M&O) costs must be identified
- ACES Complex
- Compatibility with the state’s fixed wireless access (FWA) system
Proposed Capabilities
- Real-time ability to capture improper online benefit application submittals, use of stolen identities, questionable EBT benefit usage patterns and identify potential retail store locations prone to have criminal actors utilizing stolen electronic benefit cards.
- Direct connection to EBT card vendors data to identify anomalies in criminal actor's use of the automated call center and web applications to validate EBT card balances.
- Utilize AI to identify questionable EBT terminals to gain improper access to public benefits.
- Ability for the system users to enhance fact pattern and anomaly detection methodology to stay current with the evolving methods criminal actors use in attempting to defraud public benefit systems. The system must stay nimble in a real-time environment without significant delays in change/enhancement development timelines.
- DSHS has developed algorithms, and DSHS currently utilizes data tables within a SQL database and Microsoft Excel housing analysis information to support data mining and data transformation. Tableau is used to visualize the findings for analysis and identify anomalies before taking appropriate eligibility action. Please note that providing certain information requires a formal agreement – more details and historical data of fraudulent fact patterns over approximately four years will be shared with the awarded vendors.
- Open source is preferred
- API integration is preferred
- Cloud based solutions are preferred
- Software as a solution is preferred
- Low code/No Code is preferred
- Both State RAMP and FedRAMP authorized solutions are eligible and preferred
- Currently, DSHS contracts with multiple vendors to perform written translation and on-demand in-person and over-the-phone interactions.
- Integration with the homegrown system Barcode. The solution will not replace Barcode.