The Government Can’t Cross-Check Who’s Eligible — and That Broken Plumbing Is Bleeding $186 Billion a Year.

Every year the federal government sends out tens of billions of dollars it never should have paid — to dead people, to barred contractors, to applicants who weren’t eligible. On June 16, 2026, the Government Accountability Office (GAO) told Congress one of the biggest reasons why: the federal data systems that are supposed to flag those recipients can’t talk to each other.

In a report titled Federal Data: Congressional Action Needed to Improve Interoperability of Award and Payment Eligibility Data, GAO found that agencies can tap more than 100 separate federal data sources to check whether someone qualifies for a federal award or payment — but those sources use inconsistent definitions, carry missing and duplicate records, and were never built to be cross-matched. The result is a sprawling, fragmented system in which an agency cutting a check often has no reliable way to learn what another agency already knows.

The stakes are not abstract. GAO estimates the government made roughly $186,000,000,000 in improper payments in fiscal year 2025 alone, and about $3,000,000,000,000 cumulatively since 2003. The watchdog’s recommendation to Congress is blunt: pick a single agency — it suggests Treasury — and give it the explicit authority to force these systems to work together.

• $186,000,000,000 — estimated improper payments across 64 federal programs in fiscal year 2025 — up $24 billion from the year before · Source: GAO-26-108694
• $3,000,000,000,000 — cumulative estimated improper payments since fiscal 2003 — and GAO says the real figure is likely higher · Source: GAO; Nextgov/FCW
• 100+ data sources — separate federal data sources agencies can use to verify eligibility — with no government-wide interoperability standard · Source: GAO-26-107466
• 9 of 9 — eligibility data sources GAO examined that had data-quality problems — missing, invalid, or duplicate records · Source: GAO-26-107466
• $233B–$521B — GAO's separate estimate of total annual federal losses to fraud (2018–2022 data) — the broader problem improper payments feed · Source: GAO-24-105833

GAO’s core finding is a plumbing problem dressed up in technical language. Agencies have access to more than 100 federal data sources to confirm eligibility across the life of an award — from the moment an entity applies to the moment a payment goes out. But, GAO wrote, “weaknesses in data interoperability may hinder agencies’ ability to efficiently determine award and payment eligibility.” When systems can’t reliably exchange and match records, “the risk of improper awards or payments increases.” The watchdog noted the same gap also kneecaps the government’s ability to bring artificial intelligence and advanced analytics to bear — because those tools are only as good as the data feeding them.

Keep Reading

Society · DOGE Watch

The Auditor Keeps Telling the SEC, HUD, and SBA How to Stop Losing Money — and They Keep Not Doing It

Read →

“Without data interoperability, the risk of improper awards or payments increases, and the potential use of artificial intelligence and advanced analytics to assist agencies in making eligibility determinations is limited.”

GAO-26-107466 — Federal Data, June 16, 2026

The dysfunction starts with the data itself. GAO took a close look at nine of these eligibility sources and found that all nine had quality problems — “missing, invalid, and duplicate data” — and that seven of them were inconsistent with one another, including “overlap in mutually exclusive data” that isn’t supposed to coexist. Insufficient validation rules let bad records through. When GAO matched barred-recipient records in the System for Award Management (SAM) against awards recorded on USAspending.gov, it turned up 2,074 awards to recipients who were listed as excluded at the time of the transaction — a measure of how easily the cross-checks fall through the cracks.

None of this is new, and that is GAO’s sharpest point. For more than 30 years, the watchdog said, a string of laws and guidance has set out general goals for federal data interoperability — but none established specific, enforceable requirements for the eligibility data agencies actually rely on. Goals without enforcement, repeated for three decades, produced exactly the patchwork GAO is now describing.

The government does have a central clearinghouse meant to catch these mistakes: Treasury’s Do Not Pay (DNP) system, a data-matching service agencies query to screen recipients before paying. As of September 2025, GAO found, just 28 of the 100-plus eligibility data sources were in DNP or designated for it — meaning the central check covers only a fraction of what agencies could be screening against. The most consequential feed is the Social Security Administration’s full Death Master File, which lists deceased Social Security number holders so agencies can stop paying the dead.

Keep Reading

Society · DOGE Watch

GAO: A $2 Trillion Jet That Can Fly Every Mission Only One Day in Four

Read →

When the data does flow, it works. Under a three-year pilot giving DNP access to SSA’s full death file — including state death records the system previously lacked — Treasury reported that the first year (2024) identified, prevented, or recovered $113,500,000 in improper payments against $4,600,000 in costs, a return of roughly 23-to-1. Across all of fiscal year 2025, Treasury says DNP helped agencies prevent, detect, or recover $11,700,000,000 in potential fraud and improper payments. On February 10, 2026, President Donald Trump (R) signed the bipartisan Ending Improper Payments to Deceased People Act, making the full-death-file access permanent.

To grasp why interoperability matters, look at what the leak costs. In fiscal year 2025, 15 federal agencies reported about $186,000,000,000 in improper payments across 64 programs — roughly 82% of it overpayments — up $24 billion from the prior year. The fiscal 2024 figure was about $162,000,000,000, and GAO has long warned the reported totals understate the problem because some high-risk programs don’t even produce estimates. Stack the annual numbers since 2003 and the cumulative estimate reaches roughly $3,000,000,000,000.

Most of the damage concentrates in a handful of programs. In recent estimates, the bulk of improper payments came from Medicare, Medicaid, the Earned Income Tax Credit, and the Supplemental Nutrition Assistance Program (SNAP) — exactly the kinds of high-volume eligibility programs where a missed cross-check between databases turns into a wrongful payment. And improper payments are only the visible slice: GAO’s separate, first-of-its-kind estimate pegged total annual federal losses to fraud at between $233,000,000,000 and $521,000,000,000 a year, based on 2018–2022 data.

GAO’s prescription is specific. Because three decades of general goals produced no enforceable standard, the watchdog says Congress should assign a single agency — it points to Treasury — explicit authority to lead the development of government-wide data standards and interoperability requirements for recipient eligibility data. That lead agency would coordinate with the Chief Data Officer Council and the Office of Management and Budget to set mandatory definitions, data dictionaries, validation rules, and disclosure requirements across federal systems. GAO logged this as a “matter for congressional consideration,” and as of the report’s release its status was open — the ball is in Congress’s court.

“Congress should consider assigning a single agency, such as Treasury, explicit authority to lead the development of government-wide data standards and interoperability requirements for recipient eligibility data.”

GAO's matter for congressional consideration — GAO-26-107466

Strip away the jargon and the story is simple: the federal government keeps paying money it shouldn’t because its left hand can’t check what its right hand knows. The data exists — in more than 100 sources — but it doesn’t connect, isn’t clean, and has never been forced to. Where the data does flow, as with the Death Master File feed into Do Not Pay, the savings are real and the return on investment is dramatic. GAO has now told Congress exactly what to do and which agency to hand it to. Whether lawmakers act — or let the “matter for congressional consideration” sit open while another $186 billion leaks out next year — is the part no auditor can fix. We’ll track whether Congress assigns the authority GAO is asking for.

More From Civic Intelligence

Society · DOGE Watch

GAO Puts a Number on the DOGE Cuts: Nearly 378,000 Federal Separations in 2025

The Government Accountability Office reported nearly 378,000 separations across major federal…

Society · DOGE Watch

The U.S. Spent Over $1 Billion to Counter China — and Never Checked Whether Any of It Worked

A June 2026 GAO report found the State Department and USAID spent nearly $1.2 billion on roughly 470…

Last updated June 22, 2026