Do Not Pay: How Congress Can Prevent Fraud Before It Happens
Key Findings
- Federal programs are highly susceptible to fraud and improper payments, which cost as much as $521 billion each year.
- Treasury's Do Not Pay system could be successful at stopping fraud, but only four percent of federal programs use all available data.
- The Trump administration is expanding the use of Do Not Pay as part of the crackdown on fraud and caught more than $11.7 billion in improper payments in 2025.
- Expanding the use of Do Not Pay to cover all federal programs and incorporate available data systems would prevent billions of dollars in improper payments.
Overview
Congress created the Treasury Department’s Do Not Pay Initiative to flag outgoing payments that may be improper or going to fraudulent actors. The system has three goals: prevent improper payments, detect improper payments that have already occurred, and recover misspent funds. However, Do Not Pay has not been fully utilized due to gaps in data availability and reluctance from agencies to use the system. Improper payments, which are most commonly overpayments, are a persistent issue with federal programs at nearly every agency, and are estimated to have totaled nearly $3 trillion since 2003. This is likely a significant undercount of improper payments due to pauses on eligibility reviews and improper payment tracking during this time.
When it is used by agencies and reliable data is available, Do Not Pay can both prevent and detect improper payments. The Trump administration has taken steps to enhance Do Not Pay’s utility as part of the government-wide war on fraud. Congress must act to solidify these enhancements and ensure that Do Not Pay uses all available data sources and is used by every federal program.
Rampant fraud in federal programs costs trillions
The federal government spends more than $7 trillion each year across dozens of agencies and thousands of programs. Many of these programs that disburse payments lack internal controls to prevent improper payments and fraud.
In 2025 alone, official reported improper payments totaled more than $186 billion. The true total is much higher as only three percent of the more than 2,200 federal programs report data on improper payments. Just five programs accounted for nearly 75 percent of all reported improper spending: Medicaid, Medicare, food stamps, the Earned Income Tax Credit, and a grant program for businesses impacted by the COVID-19 pandemic. The true improper payment rate for these programs is likely much higher than the official reports. Since the COVID-19 pandemic, Medicaid eligibility reviews that would flag improper payments have been suspended in some states, leading to artificially low reported error rates. In food stamps, errors under a certain monthly threshold are excluded from the calculation of improper payments.
Many instances of improper payments are due to failures by agencies to adequately confirm eligibility for a program. Fraud occurs when an individual intentionally collects payments or benefits that they are not eligible for using deception or a stolen identity. The Government Accountability Office estimates that up to $521 billion was lost to fraud each year between 2018 and 2022. This is seven percent of all federal spending during this time.
Several factors increase the risk of fraud, and welfare programs have them all. These include using self-attestation for eligibility, not regularly verifying documentation of eligibility criteria post-enrollment, large volumes of payments made, and state and local government administration of benefits.
Most fraud and improper payments are preventable, and agencies must do more to effectively steward taxpayer dollars. Do Not Pay was created to help agencies ensure payment integrity but has not delivered on that promise.
Do Not Pay can curb fraudulent spending, but only if agencies use it
Payment integrity issues have been a concern in federal programs for decades. In 2012, Congress passed legislation that created the Do Not Pay Initiative as a way for agencies to screen outgoing payments against data sources that can flag suspicious actors or payment amounts. In 2019, further legislation mandated that agencies review all outgoing payments using Do Not Pay, which functions as a centralized data hub against which agencies can screen payments, and is accessible by all federal agencies as well as state and local governments that distribute federally funded benefits and payments.
However, noncompliance with federal payment integrity requirements is widespread. In 2024, only four percent of federal programs complied with the requirement to access all data in Do Not Pay. In the same year, nearly three-fourths of all overpayments related to a data issue were made due to an agency’s failure to access existing data that would have flagged the payment as improper.
For example, an audit of just one Small Business Administration program found more than 1,600 loans made to entities on the Do Not Pay list at a cost of more than $145 million. Even though the program’s applications were cross-checked with Do Not Pay, the agency failed to take any action when a match was returned, and the payments were disbursed to ineligible recipients. A 2019 audit of the Department of Housing and Urban Development’s rental assistance program found that local governments’ failure to access Do Not Pay led to nearly $20 million in benefits paid to those who were not eligible.
Even when agencies do attempt to use Do Not Pay, they often report that it does not contain the data that is needed to ensure payment accuracy. Of the agencies that did make use of Do Not Pay in 2024, 30 percent reported that it was not effective.
Correctly implemented and used, Do Not Pay should stop fraud before it happens and help recover misspent funds. But historically, limited data access and slow adoption by federal agencies have allowed improper payments to continue largely unchecked.
The Trump administration has bolstered Do Not Pay
The Trump administration has made cracking down on waste, fraud, and abuse a top priority from Day One. Early in his second term, President Trump issued an executive order that required agencies to make greater use of Do Not Pay. This order directed agencies and the Treasury Department to use Do Not Pay to verify payments before they are sent, and to reduce administrative barriers that prevent agencies from sharing data.
The Office of Management and Budget and the Treasury Department have worked to increase agency use of Do Not Pay, and speed up the process to allow data to be shared with Do Not Pay, bringing the timeline from more than a year to just a few months. The Treasury Department has incorporated artificial intelligence and machine learning into Do Not Pay to identify and test new data sets that could have prevented an estimated $28 billion in improper payments. Do Not Pay is also now being used to detect welfare benefits being awarded to the same individuals in multiple states, identifying $1.3 billion in duplicate benefits in 2025 alone.
The improvements to Do Not Pay have netted clear results. In 2023, Do Not Pay was used to find just $652.7 million in improper payments. In 2025, Do Not Pay prevented, detected, or recovered nearly $12 billion, an 18-fold increase in effectiveness from 2023. In 2025, there were 614 million individual data pings against Do Not Pay’s databases, double the amount from 2024.
While incomplete data integration has been a key factor holding back the effectiveness of Do Not Pay, Congress and the Trump administration have made progress on this front as well. A pilot program giving Do Not Pay access to the Social Security Administration’s Death Master File saved $113.5 million in its first year alone, and is estimated to result in more than $337 million in savings over three years. The ability of Do Not Pay to access the Death Master File was set to expire this year, but Congress passed legislation making the access permanent.
Congress should build on these successes to make Do Not Pay more effective. Lawmakers have recognized the need to improve Do Not Pay and have introduced legislation to expand data integration, enhance agency use, and bolster state government access.
Congress should expand and improve Do Not Pay to prevent improper payments
Integrate all available databases
One of the factors preventing Do Not Pay from catching more fraud and improper payments is that many relevant data sources are not connected to the system. Congress should mandate that additional data sources that are already maintained by the government are integrated into Do Not Pay. These include the IRS’s Employer Identification Number data and certain other tax-related data from the IRS such as whether an individual has reported identity theft and income data that would indicate eligibility for benefits.
The Social Security Administration has finalized rulemaking that shares the Social Security Number validation service, called Numident, with Do Not Pay. Congress should codify this access to help prevent duplicate enrollment in programs, benefits sent to individuals with invalid or mismatched Social Security numbers, benefits paid to deceased individuals, and other types of improper payments related to identity verification failures.
Additionally, Do Not Pay should be linked to the National Directory of New Hires, which lists newly hired and rehired employees and can provide nearly real-time data on whether an individual is employed. Congress should also allow Do Not Pay to access data related to identity including name, aliases, and past and current addresses collected by credit reporting aggregators under the Fair Credit Reporting Act for the purpose of identity verification.
Ensure use by all federal programs
All federal programs are required by law to screen payments against the databases contained by Do Not Pay. However, only four percent of programs make use of all of the available data. Congress should require that all programs use Do Not Pay, impose consequences for noncompliance, and require regular audits of agencies’ use of Do Not Pay.
Currently, agencies often use a “pay and chase” model wherein payments that are flagged as potentially improper by Do Not Pay are sent, and agencies later make an attempt to recover overpayments. This costs taxpayers twice—first when the improper payment is sent, and again when the agency expends resources to recover the funds.
To shift to a prevention model of payment integrity, agencies should be prohibited from sending payments that have been flagged by Do Not Pay unless further investigation demonstrates that the payment is accurate. Additionally, the Treasury Department should be given the authority to stop outgoing payments that have been matched to Do Not Pay and are potentially improper rather than relying on agencies to comply.
Mandate state and local use
Medicaid and food stamps are consistently among the programs with the highest rates of fraud and improper payments, and both are administered by state agencies. Other state and local government-administered programs such as Temporary Assistance for Needy Families and rental assistance programs are also high-risk for improper payments and fraud.
State and local governments are technically authorized to access Do Not Pay to administer federally funded programs. However, screening payments against Do Not Pay is voluntary and most state and local governments do not make full use of the system due to bureaucratic processes and administrative hurdles.
Congress should require state and local governments to use all available data in Do Not Pay to verify that payments are not fraudulent or improper before they are sent out.
Together, these changes to bolster Do Not Pay would save more than $95 billion.
The Bottom Line
Congress should ensure all federal databases are connected to Do Not Pay, and that all federal programs use it.
Improper payments and fraud are a persistent threat to taxpayer-funded programs. Every dollar wasted on fraud and errors is a dollar not being used to help the truly needy, and consistent improper payments in federal programs have a significant impact on the growing national debt. For more than a decade, Do Not Pay has existed as a tool without real teeth, and improper payment rates have only worsened over time, rather than improving.
Congress should build on the Trump administration’s efforts to enhance Do Not Pay by authorizing more data integration and mandating that all federal payments are checked against all available data in Do Not Pay before a single payment goes out the door.
DOWNLOAD PAPER [PDF]