DOGE vs. Medicaid: What $1.09 Trillion in Billing Data Actually Shows

DOGE Wants You to Find Medicaid Fraud. We Already Did.

On Valentine's Day 2026, Elon Musk announced via X that the Department of Government Efficiency (DOGE) had released a “huge trove” of Medicaid spending data, inviting the American public to comb through it and find fraud themselves. Axios reported the release as a major transparency initiative. The message was clear: the government has been sitting on this data, and now it's your turn.

Here's the thing: we've already been doing this. OpenMedicaid has analyzed every publicly available CMS billing record — 227 million line items representing $1.09 trillion in Medicaid spending across 617,000+ providers. We built machine learning models, statistical anomaly detection, and network analysis tools to surface fraud signals. All from the same CMS data that DOGE is now calling revolutionary.

We applaud the transparency. More eyes on government spending is always good. But the real question isn't whether this data should be public — it's why it took DOGE to make it accessible. CMS had these records all along.

The Scale of the Problem: $37.4 Billion in Improper Payments

The CMS FY2025 improper payment rate for Medicaid is 6.12%, totaling approximately $37.39 billion. That's up from 5.09% in FY2024 — the rate is going in the wrong direction. “Improper payments” doesn't necessarily mean fraud (it includes paperwork errors, eligibility issues, and overpayments), but it's a staggering number that suggests systemic oversight failure.

To put that in perspective: $37.4 billion in annual improper payments is more than the entire budget of most federal agencies. It's not a bug in the system — it's a feature of a program that processes hundreds of millions of claims with minimal verification infrastructure.

The Fraud Is Real — And the Numbers Prove It

In 2025, the DOJ announced the largest healthcare fraud takedown in history: 324 defendants charged with $14.6 billion in intended losses. That includes everything from fake clinics billing for services never rendered to organized prescription drug diversion networks. CMS Administrator Dr. Mehmet Oz has been vocal on social media about specific cases — highlighting Minnesota's estimated $9 billion fraud problem and fraud rings in California.

Our own analysis backs this up. Across 227 million records, we've flagged 1,860 providers billing a combined $226.2 billion using multiple independent detection methods — statistical anomalies, impossible billing volumes, Benford's Law violations, and sudden behavior changes. More alarmingly, we found 40 providers actively billing Medicaid while on the federal OIG exclusion list. That's not a gray area — billing while federally banned is a crime.

Minnesota alone is a case study in how fraud metastasizes. The DOJ created a dedicated strike force for a single state. The estimated $9 billion in fraud spans home health, personal care, transportation, and interpreter services — programs where oversight is weakest and billing is hardest to verify.

$880 Billion in Cuts: The Wrong Way to Fix a Real Problem

The Republican budget bill proposes roughly $880 billion in Medicaid cuts over the next decade. The administration frames this as eliminating “waste, fraud, and abuse” — and to their credit, there's plenty of each to point to. But the data tells a more complicated story.

Our analysis shows that fraud and waste are highly concentrated. The highest-confidence fraud signals cluster in specific states, specific provider types, and specific billing codes. It's not evenly distributed across the program. Blanket cuts don't target the $226.2 billion flowing through our 1,860 flagged providers — they hit the entire program, including the roughly 90 million Americans who depend on Medicaid for healthcare.

The data supports a more surgical approach: invest in better detection infrastructure (machine learning models that can flag anomalies in real time), strengthen pre-payment verification, and actually act on the fraud signals that already exist. The OIG exclusion list is public. The billing data is public. The patterns are obvious to anyone who looks. The bottleneck isn't information — it's enforcement.

What DOGE Gets Right — And What It Misses

DOGE releasing Medicaid data to the public is, unambiguously, a good thing. Transparency is the single most effective tool for accountability. When data is public, independent analysts, journalists, and watchdogs can find patterns that bureaucracies miss or ignore. That's exactly what we built OpenMedicaid to do.

But releasing data and calling it a day isn't a strategy. The data has been publicly available from CMS for years — it's just buried in sprawling datasets with minimal documentation. What's needed isn't just access; it's infrastructure for analysis. Tools that can process 227 million records. Models that can distinguish legitimate billing patterns from impossible volumes. Cross-references between billing records and exclusion lists.

That's what we've built. Every provider, every code, every dollar — searchable, sortable, and flagged by multiple independent methods. Not because the government asked us to, but because someone should have done this years ago.

The Bottom Line

The political debate around Medicaid has become binary: either the program is sacred and untouchable, or it's a trillion-dollar slush fund that needs to be gutted. The data supports neither narrative.

There is real fraud — billions of dollars of it, concentrated in specific states and provider types. There are real people who need Medicaid — roughly 90 million of them, including children, elderly adults in nursing homes, and people with disabilities. Both things are true simultaneously.

Smart reform means using data to find and stop fraud — not using fraud as a justification for blanket cuts that affect everyone. The tools exist. The data exists. The question is whether anyone in Washington is serious about using them.

In the meantime, you can explore every provider, review the watchlist, compare states, and check any provider yourself. The data DOGE just “released”? We've had it analyzed for months.