The Centers for Medicare and Medicaid Services (CMS) reported $42 billion in annual savings from using data analytics and artificial intelligence to identify suspected fraud before payments are made, while the Department of Education projects about $1.2 billion in fraud prevention through student aid screening, according to a Federal News Network report.

Agency officials discussed the efforts during a November 5 webinar hosted by NextGov. CMS oversees roughly $2 trillion in annual Medicare and Medicaid spending.

According to the report, the shift reflects a move away from the traditional “pay-and-chase” approach, in which agencies investigate improper payments after money has been disbursed. Instead, officials are using analytical models and identity-verification tools to flag suspicious activity before funds leave government accounts.

Bethany Messick, deputy director of the data analytics and systems group within CMS’s Center for Program Integrity, said the agency uses more than 300 models to continuously monitor healthcare providers and identify behavior outside normal patterns.

Analytical alerts allow investigators to examine suspicious activity and, where warranted, suspend payments while an investigation is underway. 

CMS has also established a Fraud Defense Operations Center and a Medicaid Fraud War Room to accelerate the identification of potential fraud.

Human Oversight Remains Central

Messick said analysts are still responsible for decisions affecting payments, noting that AI systems can produce inaccurate results.

“Super Intelligence [SI] can hallucinate. It can have issues, and so we’re still in a place where we still want the human responsible for making that ultimate decision before taking action,” she said.

Education Applies Real-Time Screening to Student Aid Applications

The Department of Education is using a similar preventive approach in its Federal Student Aid operations. Richard Lucas, acting chief operating officer at the department’s Office of Federal Student Aid, said a real-time fraud-detection capability introduced into the Free Application for Federal Student Aid (FAFSA) process is on track to prevent approximately $1.2 billion in fraud.

The system screens applications as students and families complete the form. Applicants flagged for additional identity verification are asked to complete further steps, while others can proceed without unnecessary delays. 

The department says the capability was developed in partnership with a financial services firm, which it has not publicly identified.

The approach is designed to identify suspicious applications earlier while allowing legitimate applicants to access financial aid with minimal disruption.

Data Governance Becomes Central to Fraud Prevention

The expanding use of analytical models across federal programs raises important questions about data quality, identity verification, model reliability and accountability. 

Agencies must bring together information from multiple sources, identify meaningful patterns and ensure that automated systems do not unfairly delay or deny legitimate payments.

For chief data officers (CDOs), these initiatives illustrate how data governance is becoming closely connected to operational outcomes. Effective fraud detection depends on reliable data, clear rules for accessing and using it, and processes that allow analysts to understand and challenge system-generated alerts.

The federal government’s growing investment in preventive analytics also makes human oversight a critical part of responsible AI deployment. 

As agencies expand these capabilities, they will need to demonstrate not only that their systems can identify suspicious activity, but also that the resulting decisions are accurate, explainable and properly reviewed.