Why the SAVE System Is a Threat to Election Integrity
Oct 2, 2026
The US Supreme Court in an unsigned September 25 ruling approved an emergency request from the Trump administration to use a deeply flawed system for vetting voter rolls. The ruling now paves the way for states to deploy it ahead of the midterms.
Yet far from securing our elections, evidence shows the SAVE system has on multiple occasions erroneously flagged naturalized citizens as being ineligible to vote.
The Systematic Alien Verification for Entitlements, or SAVE, system relies on Social Security Administration (SSA) and U.S. Citizenship and Immigration Services (USCIS) databases to verify voter eligibility.
But a detailed assessment of SAVE system functioning by the nonpartisan General Accounting Office — now officially named the Government Accountability Office (GAO) — almost a decade ago estimated that 15%-19% of initial efforts to verify an individual’s citizenship status failed to do so. That same assessment estimated that 10%-15% of secondary verification efforts involving manual searches of paper records also failed to yield a definitive answer.
Still, some 28 states — 22 of them with Republican trifectas — are eager to deploy the SAVE system ahead of November’s elections, which polling suggests could see massive gains by Democrats as President Trump’s approval rating hits record lows amid sky high fuel prices and an unpopular war with Iran.
The Arizona legislature, Texas, as well as Florida, Indiana, and Ohio all submitted amicus briefs to the Supreme Court asserting their intent to put the system in place before voting begins.
Five reasons why SAVE is unreliable
· SAVE routinely accesses an SSA database called Numident. That database contains the Social Security numbers of almost all U.S. citizens and employment-authorized immigrants. But it does not automatically update SSN’s for newly naturalized citizens because USCIS has for several decades failed to automatically share that information with SSA.
· The underlying USCIS records about who has been naturalized are not structured to allow online searches to find them all.
· Some naturalizations — those for the minor children of a parent who naturalizes — are not finalized until the parent pays $1,385 and submits additional paperwork to secure a certificate officially confirming citizenship. Understandably, not everyone has applied for one.
· Small errors such as variants or flawed transliterations of foreign-born immigrants’ names, clerical mistakes in inputting an alien registration number — assigned by the Department of Homeland Security to non-citizens — or Social Security number into a query yield non-matches that translate into non-answers.
· The software behind SAVE was only recently updated in October 2025, while continued modifications to the system were paused as a result of litigation, meaning those who gained citizenship during this period are not in the system.
A deeper dive
There is a nearly insurmountable problem that USCIS records for about 9 million immigrants naturalized before 2000 are not online, meaning they cannot be accessed in real-time. Because of this, the SAVE technical documentation stresses that initial verification of citizenship may be inaccurate and that a second and even third verification process is needed.
That secondary process — which disproportionately affects millions of citizens who naturalized prior to the year 2000 — takes about 5 days. A third verification step, which requires submission of physical documentation of citizenship that many may not have retained, has been estimated to take about 20 days.
With these delays, even if a naturalized citizen has been allowed to submit a provisional ballot the election will be over before their vote can be counted.
Compounding matters further, in 2025 the Trump administration’s DOGE initiativepushed the SSA to pause a program — referred to as Enumeration Beyond Entry (EBE) — initiated a year earlier. Under that program, USCIS automatically notified the SSA when it approved a naturalization application and granted an immigrant citizenship.
The administration gave as its rationale for pausing EBE that it was necessary to conduct a “policy review.” In fact, a DOGE staffer at SSA had embarked on a media spree falsely claiming that “5+ million illegals” had gotten into the country to access benefits via the program.
A later GAO investigation found that automatic updates in the program only cost the federal government about $8, while in-person visits to SSA cost $55.80 per encounter.
Nonetheless, the pause remains in effect.
In the interim, estimates from the Niskanen Center, a leading think-tank, suggest more than 750,000 legal permanent residents (LPRs) have been naturalized in the past year, a significant fraction of whom have yet to see their SSA records updated, meaning they, too, would be flagged by SAVE as ineligible to vote.
Targeting eligible voters
States, meanwhile, have already begun to target voters.
Last year, Texas reviewed the status of all 18 million registered voters in the state using SAVE. By October of that year officials had flagged 2,724 individuals as potential non-citizen voters.
In September 2026, however, state investigators had determined that 578 were bona fide naturalized citizens. Only 7 were ultimately referred for prosecution for illegally voting.
The reality is that illegal voting by non-citizens, long ballyhooed by Trump and his supporters, is of such a minuscule proportion as to be a non-issue. Even the conservative Heritage Foundation has only been able to identify 68 cases of non-citizen voting going back as far as the 1980s.
The real threat to our elections comes from a White House bent on denying access to the ballot to millions of legal voters, including those newly naturalized, many of whom reside in states like Texas that will be key to determining control of Congress and, by extension, the final two years of Trump’s presidency come November 3.
Edward Kissam has conducted and published research on migrant and seasonal farmworkers and immigrant integration in the U.S., as well as educational initatives in Afghanistan. Edward has worked on issues of differential census undercount for three decades and published papers on farmworker undercount and coverage measurement in the 1990, 2000, and 2010 decennial censuses.
Comments