DOGE Medicaid Fraud Findings: What Federal Investigators Actually Uncovered
Since February 2025, DOGE operatives have had access to CMS payment systems — Medicare and Medicaid's financial backbone. They released Medicaid spending data to the public, pushed for AI-powered fraud detection, and triggered federal payment holds in multiple states. Here's what they found, what they got right, and where the investigation stands in mid-2026.
The Timeline
DOGE aides access CMS Medicare and Medicaid payment systems. Wall Street Journal reports team is “searching for fraud.”
Musk's team accesses Medicare and Medicaid records directly, bypassing normal oversight channels.
DOGE_HHS announces public release of Medicaid Provider Spending dataset — 227M billing records on HHS Open Data.
HHS defers $350M in federal Medicaid funding to Minnesota, citing fraud concerns — particularly in interpreter and personal care services.
GAO identifies $186B in total improper payments across all federal programs. Medicaid accounts for $37.4B — roughly 20% of the total.
What DOGE Got Right
Whatever you think about DOGE's methods or broader mission, several of their Medicaid-related actions were long overdue:
1. Releasing the Data
The Medicaid Provider Spending dataset — 227 million billing records covering $1.09 trillion in payments — had been sitting inside HHS for years. Making it publicly searchable was the right call. Sunlight is the best disinfectant, and crowdsourced analysis (including ours) has already identified patterns that traditional oversight missed. Our analysis flagged 1,860 providers across 9 statistical tests and ML models — including 40 providers billing while excluded from federal programs.
2. The Minnesota Crackdown
HHS deferred $350 million in federal Medicaid payments to Minnesota, targeting systemic fraud in interpreter services and personal care assistance. Our own data shows Minnesota has 4× its expected fraud rate per capita, with 81% of all interpreter fraud nationally concentrated in one state. This wasn't a political hit — the data backs it up. Minnesota was a documented outlier long before DOGE existed.
3. Pushing AI-Powered Detection
DOGE pushed CMS to adopt AI and machine learning for fraud detection — moving beyond the reactive “pay and chase” model that has defined Medicaid integrity for decades. CMS has historically relied on post-payment audits that recover pennies on the dollar. Prepayment AI screening could prevent billions in waste before checks are cut.
What DOGE Got Wrong — Or Overstated
Not every DOGE claim held up under scrutiny. Honest accountability requires acknowledging that:
- Many fraud claims weren't verified. DOGE made sweeping public announcements about fraud that, upon investigation, turned out to be coding errors, legitimate billing variations, or misunderstandings of how Medicaid data works. Statistical anomalies aren't proof — they're leads that require investigation.
- Savings claims were inflated. GAO's own analysis found that DOGE's claimed savings across all agencies didn't materialize at the scale advertised. The $186B in improper payments GAO identified is real — but DOGE didn't cause those findings, and many represent documentation errors rather than fraud.
- Access without oversight is problematic. Having DOGE operatives directly access CMS payment systems without normal audit controls raised legitimate concerns — even if the goal (finding fraud) was valid. The process matters, not just the outcome.
The $37.4 Billion Question
Medicaid's improper payment rate hit 6.12% in FY2025 — up from 5.09% the previous year. That's $37.4 billion paid incorrectly. Most of this isn't “fraud” in the criminal sense — it's a mix of:
- Paperwork and documentation failures
- Eligibility determination errors (paying for people who don't qualify)
- Billing code mistakes
- Duplicate payments
- Actual fraud by providers gaming the system
But here's the thing: the error rate is getting worse, not better. A program that can't accurately pay $37 billion worth of claims has a systemic integrity problem — regardless of whether you call it fraud, waste, or just incompetence. DOGE at least put a spotlight on a problem that previous administrations were content to ignore.
How Our Data Compares
We've been analyzing the same 227 million records DOGE released — and our findings largely corroborate the patterns they flagged, while providing more nuanced analysis:
Our Findings
- 1,860 providers flagged by 9 risk tests + ML
- 40 providers billing while federally excluded
- $37.4B in improper payments confirmed
- Minnesota: 4× expected fraud rate
- Arizona: 46 new providers, $800M+ in billing
What's Different
- We distinguish statistical flags from proven fraud
- 9 independent tests reduce false positives
- Provider-level detail, not just state aggregates
- Benford analysis, billing similarity, volume tests
- ML ensemble with feature importance reporting
Where Things Stand (Mid-2026)
DOGE as a formal entity has wound down, but its impact on Medicaid oversight continues:
- CMS is implementing prepayment AI screening for high-risk claims
- DOJ False Claims Act recoveries hit record levels in FY2025, with Medicaid fraud a priority
- Minnesota's $350M payment hold remains partially in effect pending state reforms
- The reconciliation law's work requirements (effective Jan 2027) will force the largest Medicaid eligibility verification effort in history
- HHS Open Data continues to publish updated spending datasets — the transparency DOGE pushed for is permanent
Bottom Line
DOGE's Medicaid investigation was messy, sometimes overstated, and procedurally questionable. But the core finding — that a $900+ billion program with a worsening error rate desperately needed scrutiny — was correct. The data transparency, AI detection push, and state-level accountability measures they catalyzed represent the most meaningful Medicaid integrity reforms in years. The question going forward is whether these reforms survive past the political moment that created them.
Frequently Asked Questions
What did DOGE find in Medicaid data?
DOGE investigators accessed CMS payment systems and flagged billions in suspicious spending, triggering $350 million in withheld payments to Minnesota and identifying patterns of systematic overbilling.
Did DOGE's Medicaid investigation lead to action?
Yes — their findings contributed to payment holds on states with high fraud indicators, accelerated existing OIG investigations, and pushed CMS to implement stricter pre-payment verification.
How does DOGE's approach differ from traditional fraud detection?
DOGE used data-driven analysis similar to our methodology — statistical anomaly detection and cross-referencing billing patterns — rather than relying solely on tips and manual audits.