SAVE America Act noncitizen voting claims lack audit support

The central evidentiary question in a fact check of SAVE America Act noncitizen voting claims is not whether a database can produce a large list. It is whether that list survives investigation as confirmed unlawful registration, confirmed voting, or criminal conviction. On the completed audit record available here, confirmed noncitizen voting appears at rates below 0.001% of ballots cast.

Completed audits: what survived review

The most useful evidence is not the largest allegation. It is the endpoint of an investigation cycle: the point at which election officials have moved from a record match to a reviewed registration, a verified vote, or a prosecution-quality finding. Across the available state audit materials, the confirmed numbers stay small, and the distinction between “registered,” “voted,” and “convicted” does most of the work.

Those entries are not identical evidence types. Utah’s result is an administrative audit with zero confirmed votes. Georgia’s and Michigan’s findings include confirmed voting. Louisiana’s disclosed vote count stretches back decades. The Heritage database counts convictions, not audit flags. Treating them as interchangeable would be sloppy. But they point in the same direction on the question that matters for the SAVE America Act’s factual premise: the completed records do not show noncitizen voting at a scale approaching widespread election contamination.

Michigan is the cleanest denominator in the set because it reports 16 noncitizen votes out of 7.2 million ballots in 2024, or 0.00028%. That is not a policy argument by itself. It is a measurement. If the proposed federal cure is universal documentary proof of citizenship at registration, the supporting record should show a problem broad enough to justify a universal document screen. These audits do not.

Why raw database figures inflate the claim

The recurring error is to quote the top of the funnel as if it were the bottom. A database match can be a legitimate lead. It can also be stale, duplicated, mis-keyed, legally outdated, or explained by naturalization after an earlier noncitizen record. The audit question is what remains after election officials compare records, contact agencies, check status, and identify whether a person actually voted.

Florida’s 2012 review is the caution label. The initial figure was approximately 180,000 potential noncitizens. After review, the confirmed number fell to 85. The 180,000 figure was not made up; it was a database-derived lead set. But it was not a confirmed-voter count, and using it that way would misstate the evidentiary status of the records. [6]

Iowa shows the same chain-of-custody problem in a more recent audit sequence: 2,176 initial flags became 277 confirmed noncitizens, and then 35 votes. Each step answered a different question. The first number identified records needing review. The second concerned citizenship status. The third concerned actual ballots cast. Only the last number speaks directly to the claim that noncitizens voted. [7]

That is why the July 2026 Department of Homeland Security claim involving “250,000 noncitizens in four states” should be treated as unresolved, not as proved. The figure may become meaningful if state election officials disclose the matching method, investigate the records, separate lawful naturalized citizens from current noncitizens, and report confirmed registrations and confirmed votes. As stated in the available materials, its methodology is undisclosed and no completed state investigation has verified it. [8]

That does not require assuming bad faith by the agency or by lawmakers citing the figure. It requires keeping the label attached to the evidence. A preliminary match is a lead. It is not an adjudicated violation, and it is not an audit-confirmed voting rate.

Existing law already attaches serious consequences

The SAVE America Act debate is sometimes framed as if noncitizen voting exists in a legal vacuum. It does not. Federal law already reaches the documentation and immigration consequences tied to unlawful registration and voting conduct. Section 1324c addresses document fraud, and Section 1227(a)(3) includes deportability consequences for covered violations. [9][10]

That context does not end the policy debate. Congress can decide that existing enforcement is insufficient. But if the new requirement is defended as a response to widespread noncitizen voting, the proof should show widespread voting, not merely the legal possibility of abuse or the existence of already punishable conduct.

The access-burden claim is separate from the audit claim

There is also a burden side to the proof-of-citizenship proposal. The Brennan Center has estimated that 21.3 million U.S. citizens lack ready access to documentary proof of citizenship. That figure should be read with attribution: it is a Brennan Center estimate, not a completed government audit of every affected voter. [11]

Still, the number matters in a regulatory proportionality analysis. A universal document requirement does not fall only on the small class of confirmed violators. It changes the registration process for eligible citizens as well. The audit record, therefore, has to do more than show that noncitizen voting is illegal or that some cases exist. It has to support the scale of the proposed screen.

Legislative status, verified in Q3 2026

As of 2026-07-28 UTC, the SAVE America Act language was moving through Congress in fast-changing form. The measure had been attached to the National Defense Authorization Act, passed the House 216-212 on July 22, 2026, and was in the Senate. [12]

That status explains the urgency of the fact check, but it does not change the evidentiary standard. A live bill can create pressure to compress allegations into talking points. It does not convert database flags into confirmed votes.

Narrow conclusion

The available completed state audits do not support the claim of widespread noncitizen voting behind a universal documentary-proof-of-citizenship requirement. They show isolated confirmed registrations, smaller confirmed vote counts, and—in Michigan’s reported same-election figure—a confirmed rate of 0.00028% of ballots cast.

The strongest contrary numbers remain preliminary database claims until they pass through state investigation and are reported as verified registrations, verified votes, or convictions. On the present record, the SAVE America Act’s factual premise is not audit-confirmed.

References

1. Utah Lt. Gov. Henderson announcement, Utah Lieutenant Governor
2. Georgia Secretary of State audit, Georgia Secretary of State
3. Louisiana elections disclosure, Louisiana elections officials
4. Michigan Secretary of State Benson report, Michigan Secretary of State
5. Election Fraud Database, Heritage Foundation
6. Florida DHSMV review, Florida Department of Highway Safety and Motor Vehicles, 2012
7. Iowa Secretary of State Pate audit, Iowa Secretary of State
8. DHS “250,000 noncitizens in four states” claim, Department of Homeland Security, July 2026
9. 8 U.S.C. § 1324c, United States Code
10. 8 U.S.C. § 1227(a)(3), United States Code
11. Brennan Center estimate on citizens without proof of citizenship documents, Brennan Center for Justice
12. SAVE America Act legislative status, Congress, July 22, 2026