California Ballots Saved From DOJ Dragnet

Hand casting ballot into a box at a polling station — Photo: Sangiao Photography / Shutterstock

A federal judge’s rejection of the Justice Department’s bid to isolate 20,000 California mail ballots turned on a question that will outlast this particular election cycle: how much confidence should courts extend to an unverified administrative list when the remedy sought is withholding votes that have already been cast.

Key Points

The DOJ sued California days before the November election, claiming Homeland Security Investigations identified roughly 20,000 “likely noncitizens” registered to vote and asked the court to segregate their mail ballots pending citizenship verification.
U.S. District Judge David Carter denied the request, finding the government had not explained its matching methodology and that the proposed remedy risked disenfranchising eligible voters.
A Los Angeles County sample of 200 names from the list found nearly 40% indicated they were U.S.-born and 71.2% had no voting history at all, undercutting the list’s reliability.
The Ninth Circuit declined to intervene on an emergency basis, ruling the district court hadn’t been given adequate time to decide the matter first.
National data on noncitizen voting, including a Brennan Center survey of 23.5 million ballots, has consistently found the practice exceptionally rare, framing this dispute as one about data quality, not a demonstrated wave of illegal votes.

What the Justice Department Actually Alleged

On October 15, the DOJ filed suit against California, asserting that Homeland Security Investigations had compiled a list of approximately 20,000 people it “reasonably believes” are noncitizens registered to vote, and that California had mailed them ballots. The complaint invoked the uncontested legal baseline that aliens may not vote in federal elections, and it asked the court for something narrower than outright ballot rejection: a temporary order requiring California to segregate the flagged ballots and verify citizenship before tabulation. That framing — investigation and verification rather than immediate cancellation — was central to how the department defended its request in court.

The department’s own account of events matters here: concerns were reportedly flagged internally as early as mid-2023, yet the lawsuit was not filed until mid-October, after ballots had already been mailed and many returned. That timing compressed the available window for judicial review to days, not weeks, and it shaped nearly everything that followed.

Why the Court Said No

Judge Carter’s 26-page order did not dispute that noncitizen voting is illegal; it disputed whether the government had shown its list was accurate enough to justify freezing ballots six days before an election. Carter found the DOJ’s account of how it built the list “vague” and said the explanations “raised multiple questions regarding reliability”. He wrote that “once a vote is segregated, lasting harm has occurred,” and declared that “this Court declines to be a party to such process”. He further characterized the requested overhaul of California’s ballot-processing procedures as a “monumental change” imposed on the state on short notice, and noted that federal prosecutors did not need to inspect ballots to pursue illegal-voting cases, since returned envelopes already carry the voter’s signature.

That last point goes to the heart of the dispute: California’s existing procedures already generate a paper trail — a signed attestation of citizenship under penalty of perjury — that officials argued is itself evidence, and that ordinary post-election investigation and prosecution remain available without disturbing ballots still in the counting pipeline. Carter’s ruling leaned on that alternative, concluding the government hadn’t shown why a blanket administrative hold on an entire flagged population was necessary when individualized prosecution was already an option.

The Data Problem That Undercut the Case

The most consequential fact to emerge from the litigation wasn’t legal — it was empirical. Los Angeles County reviewed a sample of 200 names drawn from the DOJ’s list and reported that nearly 40% indicated they were born in the United States, and that 71.2% had no voting history whatsoever. If those proportions held across the full 20,000, a significant share of the “likely noncitizens” the DOJ sought to flag would be citizens who had never voted in the election at issue — not evidence of fraud, but evidence of a matching process generating false positives. Voting-rights expert Kassra Oskooi separately testified that the USCIS matching protocol underlying the list was “likely to be unreliable”, and state officials noted the original claim had already shrunk once, from an initial estimate near 190,000 names down to 20,000, after closer scrutiny.

California Secretary of State Shirley Weber and Los Angeles County Registrar Dean Logan added a procedural complaint of their own: the DOJ transmitted the list on a password-protected drive without accompanying data or methodology, and Logan declined to open it over security concerns before litigation even began. California Attorney General Rob Bonta went further rhetorically, calling the list “unreliable” and “inaccurate” and pointing to the broader evidentiary record — election counts, recounts, hand counts, and audits — that he said consistently shows voter fraud is “exceedingly rare”. That claim is not merely advocacy; it tracks a 2017 Brennan Center survey of 42 jurisdictions covering 23.5 million ballots, which found roughly 30 suspected noncitizen-voting cases referred for investigation — about 0.0001% of votes cast. None of this proves zero noncitizen voting occurred in California in this cycle. It does establish that the burden of proof the DOJ carried into court was heavier than its public list could bear.

The Appellate Dead End

The Justice Department sought emergency intervention from the Ninth Circuit before Judge Carter even ruled, trying to accelerate a remedy ahead of Election Day. The appellate panel declined, concluding the district court hadn’t been given adequate time to decide the matter in the first instance. That procedural rebuff mattered less for its legal substance than for its signal: an appellate court was unwilling to bypass ordinary fact-finding on a record this contested, even with an election deadline bearing down. It left the DOJ’s broader lawsuit alive but its emergency leverage spent, with no pre-election mechanism left to compel California’s compliance.

What This Dispute Actually Resolves — and What It Doesn’t

Nothing in Judge Carter’s ruling forecloses the DOJ’s underlying lawsuit, and nothing in it suggests noncitizen voting is categorically impossible or unworthy of pursuit; federal law on that point is unambiguous. What the ruling forecloses is a specific tactic: using an unvalidated administrative match, compiled without disclosed methodology, to freeze ballots belonging to tens of thousands of people in the campaign’s final days. The pattern here — an initial number that shrinks on scrutiny, a sample review finding a large share of flagged individuals are citizens or nonvoters, and a court unwilling to let unverified data override state election procedure — has recurred in other jurisdictions facing similar DOJ list demands, including a widely cited Nevada case where an initial estimate of roughly 15,900 potential noncitizens collapsed to a small fraction after verification. The durable lesson for future election cycles is that enforcing citizenship requirements and protecting against list-driven disenfranchisement are not opposing goals — they both depend on the same thing the DOJ’s list lacked here: a transparent, auditable methodology produced well before ballots are in voters’ hands, not after.