Executive Summary
The U.S. Government Accountability Office reported $162 billion in improper payments across 68 federal programs in fiscal year 2024 — representing a fraction of actual fraud. Since 2003, cumulative improper payments have exceeded $2.8 trillion. Seventy-five percent of this total concentrates in just five programs: Medicare, Medicaid, the Earned Income Tax Credit (EITC), Supplemental Nutrition Assistance Program (SNAP), and pandemic-era spending initiatives.
The root cause of this epidemic is structural: states detect fraud program-by-program, never connecting patterns across multiple systems. Fraud rings do not operate within a single program — they simultaneously exploit multiple systems. A healthcare provider committing Medicaid fraud is simultaneously operating as a SNAP-authorized retailer trafficking benefits, filing false child welfare claims, or billing workforce training programs. Current oversight approaches, relying on agency-specific OIG investigations and annual audits, catch only 5–10% of fraud vectors.
The solution is a multi-agency detection architecture that correlates data across all state-funded programs simultaneously. Sentinel Integrity Group's platform ingests data from every program, applies AI pattern detection across all of them, and surfaces cross-program fraud invisible to single-agency tools. The result: 3–5x more fraud detected, faster response times, and measurable recovery of improper payments before they escape state coffers.
The Scale of Federal Improper Payments
| Program | Improper Payments (FY2024) |
|---|---|
| Medicare | $51.0 billion |
| Medicaid | $29.0 billion |
| Earned Income Tax Credit | $22.0 billion |
| Unemployment Insurance | $19.0 billion |
| SNAP | $7.0 billion |
| Other Programs (63 additional) | $34.0 billion |
| TOTAL | $162.0 billion |
These figures represent what federal oversight and single-agency audits detect. The actual fraud is conservatively estimated at 2–3 times higher. States operate in data silos: the Medicaid fraud unit operates independently from the SNAP integrity unit, which operates independently from child welfare oversight and workforce program compliance. Federal oversight mirrors this fragmentation — the Centers for Medicare & Medicaid Services (CMS) oversees Medicaid separately from the Food and Nutrition Service (FNS) overseeing SNAP, which operates separately from the Department of Labor and Administration for Children and Families.
Cross-program fraud patterns are the norm. A behavioral health provider network billing Medicaid for phantom therapy services simultaneously operates as a SNAP-authorized retailer trafficking Electronic Benefit Transfer (EBT) benefits. The same entity may bill child welfare for fabricated supervised visitation hours and workforce programs for ghost students in training programs that never occurred. When audited in isolation, each program's billing appears plausible. When correlated across all programs, the provider's claimed activities sum to 28+ hours per day — an obvious impossibility.
The "whack-a-mole" problem emerges: shut down fraud in one program through enforcement, and perpetrators shift operations to another. A provider excluded from the Medicaid network can immediately enroll as a SNAP vendor or child welfare contractor. Without cross-program visibility, the same bad actor resurfaces, defrauding a different program. Federal recoveries under current approaches average $0.04 per improper payment dollar — unacceptable when $162 billion annually escapes into fraud.
Structural Failures of Single-Program Oversight
Four structural factors render single-program fraud detection ineffective at scale:
1. Data Isolation
Each agency maintains its own data systems with independent governance. Medicaid Management Information Systems (MMIS) exist separately from SNAP-EBT systems, which exist separately from child welfare case management systems (SACWIS), workforce systems (WIOA), and housing programs. No cross-referencing occurs across these systems. A healthcare provider excluded from Medicaid networks can immediately enroll as a SNAP-authorized vendor, a child welfare service provider, or a workforce training contractor. The exclusion information does not propagate. Without a unified entity registry linking providers across all programs, the same bad actor operates under different entities and nominally compliant registrations.
2. Detection Model Limitations
Single-program fraud detection models are trained on program-specific billing patterns and behaviors. These models cannot detect fraud when a provider's billing is individually normal within each program but collectively impossible. A healthcare provider billing Medicaid for 8 hours of therapy services, operating a SNAP retail location requiring 10 hours of daily on-site management, and billing child welfare for 6 hours of supervised visitation — all on the same day — passes every single-program edit check. Only cross-program correlation reveals the temporal and physical impossibility.
3. Audit Cycle Mismatch
Different programs audit on different schedules. Medicaid conducts post-payment reviews quarterly. SNAP performs retailer audits annually. Child welfare reviews foster care placements bi-annually. Workforce training audits occur on a three-year cycle. A fraud ring expertly exploits these timing gaps — filing false claims in November (after the most recent Medicaid review, before the next), shifting to SNAP in April (between annual audits), and cycling through workforce programs on a three-year rotation. By the time the next program-specific audit occurs, the fraudulent entries are stale and difficult to investigate.
4. Jurisdictional Gaps
Federal oversight is agency-specific. CMS oversees Medicaid compliance. FNS oversees SNAP. ACF oversees child welfare. DOL oversees workforce programs. State Offices of Inspector General often lack formal cross-program authority. No single authority is charged with watching the gaps between programs or connecting patterns across agency boundaries. Fraud operating at the intersection of multiple programs goes unseen because no single oversight entity has incentive or authority to look across all programs simultaneously.
A provider billing Medicaid for 8 hours of therapy, SNAP-authorized for retail operations requiring 10 hours of daily management, and billing child welfare for 6 hours of supervised visitation — on the same day — passes every single-program edit check. Only cross-program correlation reveals the impossibility.
A Framework for Cross-Program Integrity
The solution is a unified architecture that ingests, standardizes, and analyzes data across all state-funded programs simultaneously. This framework operates on four layers:
1. Universal Data Ingestion
Accept data from every state-funded program in any format — MMIS extracts, SNAP-EBT exports, SACWIS case data, WIOA participant records, housing program voucher data. No integrations required. Agencies upload via secure SFTP on their existing schedules. The platform normalizes disparate data schemas to a common entity model. A Medicaid provider NPI becomes linked to the same entity's SNAP vendor number and child welfare service provider registration.
2. Cross-Program Entity Resolution
Match providers, vendors, beneficiaries, and organizations across programs using probabilistic matching algorithms. The same entity billing Medicaid as "ABC Behavioral Health LLC" and operating as "ABC Community Services" in child welfare gets unified into a single profile. Suspicious variations (ABC Health Behavioral vs. Behavioral Health ABC) trigger attention. The platform maintains a living entity registry mapping every provider and vendor across all programs in real time.
3. Multi-Program Pattern Detection
AI models trained on cross-program fraud patterns, not single-program anomalies. The system flags impossible billing combinations (28 hours of claimed services per day when summed across programs), suspicious entity networks (a provider whose vendors, subcontractors, and beneficiaries appear in Medicaid, SNAP, child welfare, and workforce programs), timing impossibilities, and geographic anomalies (a provider with no physical location claiming to serve beneficiaries across five states). These patterns are invisible to single-program analysis but obvious in cross-program context.
4. Unified Oversight Dashboard
A single pane of glass for legislators, agency directors, and compliance officers. See fraud risk across ALL programs simultaneously. Drill down by program, provider, vendor, beneficiary, geographic region, or fraud category. Compare fraud risk and detection rates across programs. Identify which programs are most vulnerable. Track recovery rates and ROI. Executive dashboards support policy decisions; investigator workbenches support case development and rapid response.
Recommendations for State Legislators and Agency Directors
1. Mandate Cross-Program Data Sharing — Enact legislative proviso requiring data-sharing across all state-funded programs — not just health programs, but workforce, child welfare, housing, and transportation. Data sharing is not optional; it is foundational to fraud prevention at scale.
2. Establish Minimum Data Reporting Standards — Require every program receiving state appropriation to report standardized data elements quarterly. Define a Minimum Data Reporting Standard (MDRS) covering entity identifiers, transaction amounts, service dates, and provider information. Compliance with MDRS becomes a condition of state funding.
3. Fund Cross-Program Integrity as a Line Item — Establish a dedicated, unified integrity function reporting to the State Finance Office or Governor's office — not buried within individual agency budgets. Cross-program oversight cannot be an agency afterthought; it requires independent funding and authority.
4. Deploy Unified Entity Registries — If a provider or vendor is excluded from one program (due to fraud, abuse, or non-compliance), they are automatically flagged across all programs. Implement real-time exclusion matching so bad actors cannot simply move to the next program.
5. Deploy AI-Powered Multi-Program Detection — Technology for cross-program pattern detection exists today and costs a fraction of traditional audit and investigative approaches. Deploy AI models trained on cross-program fraud patterns. The ROI is measurable: 3–5x more fraud detected, faster response, and recovery of improper payments before they escape state systems.
About Sentinel Integrity Group
Sentinel Integrity Group, an HRN Group division, provides AI-powered fraud prevention and program integrity solutions to state and federal agencies nationwide. Our platform ingests data from Medicaid, SNAP, TANF, child welfare, workforce training, and housing programs — covering over 50 fraud vectors across 8 program integrity lifecycle stages.
Unlike single-program solutions, Sentinel's architecture detects cross-program fraud patterns invisible to traditional audit and investigative approaches. No system integrations required. Agencies upload data via secure SFTP on their existing schedules. Live monitoring and alerting begin within 2 weeks of contract signature.
Sentinel partners with state legislatures, governors' offices, and agency leadership to recover improper payments and strengthen program integrity at scale.
Contact:
David Thorne, CEO & Founder
david@highvaluechange.com
(316) 393-8324
sentinelintegritygroup.com