Watch Out for False Voter Fraud Claims Fueled by SAVE Program

This year, observers should be wary of voter fraud claims made by government officials and private groups, even if those claims cite federal government data. Many such false or misleading claims promoted over the past year have relied primarily on the flawed Systematic Alien Verification for Entitlements (SAVE) program, which this article addresses in detail.

The second Trump administration has been engaged in a concerted campaign to undermine elections. Part of that involves collecting state voter files and using federal data sources to lend pseudolegitimacy to false claims of widespread fraud. While there may be valid ways to use federal data to support election officials’ efforts to keep voter rolls accurate and up to date, there are notable shortcomings in such data, and it may be misused to spread misinformation.

To properly assess any claim of voter fraud based on federal data, it is important to understand the flaws in the data, common analytical pitfalls, and the ways faulty data may be exploited to disenfranchise eligible voters and erode confidence in elections.

In recent election cycles, private groups and a small handful of state officials have latched on to faulty data to cast doubt on the accuracy and reliability of state voter rolls and to allege widespread voter fraud. This year, our elections are facing a new threat: a federal government seeking to unlawfully insert itself into voter list maintenance. The administration has enlisted the Department of Justice to bully states into releasing sensitive voter data, and it is pushing the use of an expanded version of the SAVE program — a Department of Homeland Security tool designed to help verify the citizenship of applicants for government benefits — to identify supposed non-U.S. citizens on the voter rolls.

A March executive order now also instructs DHS to use the SAVE program and other sources of federal data to compile “citizenship lists” for states ahead of federal elections — seemingly to supplant states’ own voter lists, without any authorization under federal law to do so. It orders the DOJ to investigate and prosecute officials who allow supposed noncitizens to vote. The Brennan Center and other groups are challenging this unlawful executive order in court. The administration has admitted in a separate case challenging the executive order that such citizenship lists would likely be unreliable.

While it is not improper for the federal government to assist states in verifying voters’ eligibility through the SAVE program — indeed, past administrations have done so — this administration’s actions appear closer to taking over the task of maintaining states’ voter lists, a role that Congress has delegated exclusively to the states under its constitutional authority. The SAVE program’s underlying flaws, its hasty overhaul, and the government’s failure to provide proper guidance to state and local officials on responsible use of the expanded tool all mean that the SAVE program has produced numerous errors that falsely identify voters as noncitizens. At the same time, the administration has asked that states enter into agreements that would purport to allow it to share states’ voter information with outside groups, compounding the risk of federal data being misused to spread false claims about noncitizen voting.

Last year, the Trump administration rushed through sweeping updates to the SAVE program with the stated aim of ridding elections of the “taint” of “illegal aliens . . . defraud[ing] Americans.” Previously, the program only allowed users to search, one by one, for people who had immigration records or some other involvement in formal DHS processes — for example, individuals born overseas to American parents who applied for a certificate of citizenship. With the 2025 changes, the program allows users to conduct bulk searches for anyone with a Social Security number, enabling officials to query massive numbers of registered voters at once. The SAVE program can now also access passport data from the Department of State, according to emails between U.S. Citizenship and Immigration Services (USCIS) and a local election office, as well as interviews with local officials.

All states use data in some way to maintain voter lists. But large-scale data-matching programs always carry the risk of false positives, so registered voters flagged by such matching merit further investigation. Most election officials employ robust safeguards to protect voters from disenfranchisement, such as cross-referencing voter records against other state sources and giving voters a chance to clear up mistakes before they are removed from the rolls. But in past years, a small number of officials have tried to remove voters without taking proper precautions to verify their eligibility, jeopardizing eligible voters’ registration status and feeding unfounded doubts about election integrity. That risk is even greater in 2026 given the Trump administration’s efforts to use the flawed SAVE program to support false claims that noncitizens are voting in large numbers.  

There are three principal reasons to be skeptical of voter fraud claims based on SAVE program results.

Error-Prone Methodology at Scale

Large-scale data matching consistently produces errors in a range of contexts. Individuals often appear in multiple databases, and inconsistencies in the listing of their name (for example, Juan Gonzalez appearing as J. Gonzales in some records), birthdate, or gender may prevent a data-matching system from querying all their associated records. Large data ecosystems can also struggle to integrate new information. For example, a system may fail to recognize that Anita Das’s name has changed to Anita Das-Reilly after marriage. These gaps and discrepancies create problems in the context of voter list maintenance. Consider the case of a naturalized voter whose U.S. citizenship is recorded in a USCIS database but not in the Social Security Administration database and other government records, which could lead to them being incorrectly flagged as a noncitizen voter.

Similar mistakes are highly likely. USCIS maintains many databases that collectively contain more than 100 million immigration records, and the SAVE program can access a large amount of that information. Managing that volume of data is enormously complex, and even under the best circumstances, USCIS makes mistakes when identifying people’s citizenship or immigration status. When data reconciliation is done at such a mammoth scale, a modest error rate (at least) is inevitable. Given the number of individuals processed, however, even a small error rate could translate into tens of thousands of voters being mistakenly identified as noncitizens. 

To put that in perspective, consider that peer-reviewed social scientists have viewed false-positive error rates ranging between 0.9 percent and 2.5 percent as acceptable when drawing conclusions from large-scale data reconciliation. (There is no generally accepted false-positive error rate.) Then consider New York Times reporting from earlier this year revealing that DHS ran 49.5 million voter files through the SAVE program and seemingly identified around 10,000 registrants as potential noncitizens. That represents just 0.02 percent of the total number of registrants in the voter files searched, a percentage far lower than many typical error rates. In other words, when dealing with datasets of this size, it would not be unreasonable to expect at least 10,000 false matches. In fact, that would reflect a low error rate.

As such, the numbers the administration has disclosed could be entirely explained by errors inherent in any large-scale data-matching process. To be clear, there is no publicly available evidence that all those matches are false positives, but as discussed further below, there are plenty of indications that at least a large percentage of them likely are. The point is that even very low error rates in the SAVE program could produce a number of matches that seems quite large in absolute terms.

Like all systems that match and reconcile information across vast datasets, the SAVE program has a baseline error rate. If the administration successfully compiles a complete list of the roughly 174 million registered voters in the country and runs it against the SAVE program, that baseline error rate alone, even if small, could give the false impression that there is a massive number of ineligible or potentially ineligible voters on the rolls. Americans falsely flagged by the SAVE program could face unnecessary burdens to prove their citizenship or be disenfranchised altogether if they, for example, miss a notice from their local election official in the mail.

Faulty or Incomplete Data Sources

Beyond the matching errors that commonly afflict large data systems, the SAVE program has specific flaws that undercut its reliability.

First, the SAVE program can pull from particularly error-ridden datasets, including U.S. Customs and Border Protection’s Automated Targeting System and TECS, the primary system used to screen the admissibility of those arriving at a U.S. border. The ATS dataset incorporates information from the FBI’s terrorism watchlist, which contains unreliable information misidentifying people as terrorists and permits intentional biases against certain religious and racial minorities. TECS integrates that same FBI watchlist information, as well as travel records and other fallible data sources.   

Compounding these pitfalls are newly SAVE-accessible data sources from the Social Security Administration and the Department of State, which have their own shortcomings. The Social Security Administration only began systematically collecting citizenship data in 1978, and the agency’s notations indicating that someone is foreign-born or a noncitizen may have been inferred for Social Security number holders born before that date. The Social Security Administration’s central database often lacks up-to-date citizenship information for naturalized citizens if they did not notify the agency of their naturalization. And roughly half of American citizens lack a passport, making Department of State passport data meaningless for tens of millions of Americans.   

Second, no dataset feeding into the SAVE program is likely to contain the most up-to-date information on naturalized citizens. DHS publishes a list of hundreds of thousands of naturalized citizens each quarter. No matter when a search is run through the program, there will be substantial numbers of newly naturalized people whose records may not have been updated promptly for key datasets used by the SAVE program. Instances like these, where records reflect outdated information, illustrate why relying on the SAVE program alone for citizenship verification is unwise.   

Third, USCIS’s hasty expansion of the SAVE program was finalized before updates to citizenship data could be processed and merged, resulting in blunders, according to news reporting. The agency has admitted that it initially provided incorrect information to at least five states through the SAVE program. The program wrongly flagged hundreds of voters across Missouri as noncitizens. One county clerk in Boone County, Missouri, said that more than half of registered voters flagged by SAVE as noncitizens in November turned out to be U.S. citizens. As another election official put it, the expanded SAVE program “is not ready for prime time.”

Track Record of Incorrect Results

States or localities that have supposedly identified substantial numbers of noncitizen voters by using the overhauled SAVE program have been proven wrong. The outcomes of these investigations, the inherent pitfalls of large-scale data matching, and the flawed data sources used by the SAVE program all demand significant caution regarding any future claims about noncitizens on voter rolls.

For example, around 35 percent of registered voters identified as noncitizens in St. Louis County, Missouri, by the SAVE program were, in fact, naturalized U.S. citizens who registered to vote at their naturalization ceremonies, according to the county’s Republican election director. Likewise, after Texas officials announced that the SAVE program had identified hundreds of registered Texas voters as potential noncitizens, subsequent public reporting found that they had failed to cross-check state records confirming the registrants’ citizenship.

A number of states have recently taken a more responsible approach to using the SAVE program that underscores the pitfalls of relying exclusively on federal data sources and confirms the rarity of noncitizen voting. For instance, after Utah completed a review of its voter rolls, it found zero occurrences of noncitizen voting. Last year, Louisiana reviewed its voter records going back four decades and identified only 79 potential noncitizens who voted, out of an estimated 74 million ballots cast during that period. Louisiana’s Republican secretary of state declared that “non-citizens illegally registering or voting is not a systemic problem in Louisiana.”

In 2025, the Department of Justice sent letters to 18 states, demanding that they enter into information-sharing agreements with the agency. The text of the proposed written agreement says that voter files — which contain sensitive personal data such as full names, home addresses, dates of birth, driver’s license numbers, and partial Social Security numbers — may be provided to a contractor for work “related to the Department’s list maintenance verification procedure.” This language opens the door to third parties inappropriately gaining access to voter information from the federal government. Should that scenario bear out, any claims coming from nongovernmental entities about voter eligibility should also be met with skepticism, both because of the inherent shortcomings of the SAVE program and because similar third-party efforts have consistently proven to rely on shoddy data analysis.

Indeed, in recent election cycles, election denier organizations and associated groups of private individuals mounted vigilante efforts to challenge the registrations of thousands of supposedly ineligible voters they identified through amateurish data matching, comparing lists against public sources such as National Change of Address lists, maps, property tax records, and newspaper obituaries. Election denier groups have also filed lawsuits that have perpetuated lies about voter fraud, often basing their allegations on glaring methodological errors. These schemes have been characterized by common data-matching pitfalls, such as false matches or information gaps due to identical names or addresses, transposition errors or inconsistencies, typos, unreliable sources, and outdated address-change lists. Radical right-wing groups have seized on these slippery efforts to concoct false claims and attempt to undermine confidence in elections. It is critical for election officials and the public to be cautious in reacting and responding to claims promoted by similar groups in 2026.

• • •

The SAVE program can help election officials conduct voter list maintenance, but only if used prudently alongside established, trusted practices for maintaining accurate voter rolls. Misuse of the SAVE program is already fueling false claims that voting by non-U.S. citizens is widespread — a myth that has been thoroughly disproven, again and again. Unless election officials and the federal government use the SAVE program with great caution, eligible U.S. citizens could bear the burden of re-proving their eligibility or be disenfranchised altogether. And the public, in turn, should question fraud claims that rely on the SAVE program’s results.