The Department of Homeland Security (DHS) is urging state election officials to utilize the Systematic Alien Verification for Entitlements (SAVE) system to audit voter rolls, citing concerns that hundreds of thousands of non-citizens are registered to vote. DHS Secretary Markwayne Mullin stated that the agency claims approximately 250,000 non-citizens are registered across the U.S., with a significant portion originating from four specific states. However, the push comes as experts and some state officials have previously flagged inaccuracies within the SAVE system itself.

The SAVE system, maintained by U.S. Citizenship and Immigration Services, is designed to verify immigration and citizenship status. While historically used to determine eligibility for government benefits and restricted to querying known immigrants, its scope was expanded under the Trump administration to allow for broader checks of voter rolls, including natural-born citizens, by utilizing Social Security numbers and data from the Social Security Administration (SSA).

Concerns about the system's reliability have been raised by various entities. A fact sheet from USCIS, since removed, acknowledged that SAVE might not confirm citizenship for individuals not designated as U.S. citizens in SSA records. The SSA itself noted in 2023 that its citizenship data represents a snapshot and does not definitively confirm status. A 2006 audit of SSA data found that 7% of individuals classified as non-citizens were actually U.S. citizens who had not updated their status, which at the time translated to millions of citizens being misidentified.

The expanded SAVE system has also faced legal challenges. The U.S. District Court for the District of Columbia ruled the expansion violated several acts, including the Privacy Act, and blocked its use, stating some states were removing citizens based on inaccurate information. Conversely, a federal judge in Florida later ruled that the government must continue providing access to the expanded system for four states. The Department of Justice indicated efforts to comply with both rulings, restoring access only to states covered by the Florida order.

State officials have also reported issues. Utah's comprehensive review in 2025 using SAVE found nearly 9,000 individuals whose citizenship could not be confirmed. Subsequent manual review revealed over 5,600 of those were citizens, with another 5,000 unconfirmed and only 27 confirmed non-citizens.

Earlier this year, Texas's former secretary of state expressed concerns about the accuracy of data from the system. While county investigations identified 2,724 potential non-citizens, the state was aware of instances where state driver's license records inaccurately reflected citizenship status and clerical errors led to non-citizens being registered to vote despite indicating their status on applications.

DHS has stated that USCIS has made enhancements to SAVE and implemented manual review processes to ensure accuracy. However, officials in California and Nevada, two of the four states identified by DHS, have expressed skepticism. California Secretary of State Shirley Weber stated that the provided information did not inspire confidence in the methodology used. Nevada Secretary of State Francisco Aguilar described the numbers as "wildly speculative" and lacking substantiation.

Secretary Mullin sent letters to election officials in California, New Jersey, Pennsylvania, and Nevada, urging them to collaborate on identity verification by a July 30 deadline. The letters cited a "preliminary review" of publicly available voter registration data, claiming matches with non-citizens in DHS files based on name, date of birth, address, and Social Security number. This claim has been questioned, as publicly available voter data typically does not include Social Security numbers.

Sean Morales-Doyle of the Brennan Center for Justice called the methodology outlined in the letters dubious, stating that matching Social Security numbers against public voter data, which does not contain such information, is illogical. He described the potential results as "garbage in, garbage out" if the SAVE tool was used for cross-checking, adding that the numbers conflict with available data and history and are likely "worthless."