Fact Check: Were 449,000 California Noncitizens Registered to Vote? No Evidence Supports That

The claim and verdict

Posts circulating before the 2026 midterms say 449,000 California registered voters admitted they were noncitizens when summoned for jury duty. The conclusion is unsupported. AFP traced the closest matching figure to a Judicial Council report for fiscal year 2016–17: 449,404 of 1,096,478 disqualified prospective jurors were removed for citizenship status. The report counted jury disqualifications, not voter registrations.

The viral argument silently substitutes one database for another. Receiving a jury summons can show that a name entered a jury-source list. It does not show which source supplied the name, whether the person was registered to vote or whether the person ever cast a ballot.

Three records the post collapses into one

California courts assemble prospective-juror lists from voter registration, Department of Motor Vehicles records and state tax-filer information. Noncitizens may lawfully hold an AB 60 driver's license and may file California taxes. Either route can place an eligible resident's name in the pool used to send a questionnaire or summons even though that resident is correctly ineligible to serve on a jury.

Voter registration is a separate record. California requires a registrant to be a United States citizen and to attest to the information under penalty of perjury. Federal law also generally prohibits noncitizens from voting in federal elections. None of those safeguards proves that unlawful registration never occurs; they show why a jury-disqualification total cannot measure it.

What evidence would be needed

A defensible estimate would require a lawful, audited match between current voter files and reliable citizenship records, with duplicate names, address changes, naturalizations, deaths, data-entry errors and local-only registrations resolved. It would then have to distinguish registration from an actual ballot. The recycled 2016–17 jury statistic does none of that work.

The age of the number is another warning. A figure tied to one fiscal year cannot automatically describe the September 2026 electorate. AFP reported that the same statistic has circulated since at least 2018, often without its date or denominator. Removing those labels makes an old administrative count look like a new discovery.

A four-question test for election claims

Applied here, all four questions break the viral claim. The 449,404 figure is real in its reported context, but that context is jury eligibility. Presenting it as proof of 449,000 registered noncitizen voters is false.

• What did the original record count: jurors, registrations, ballots or prosecutions?
• Which lists fed the record, and could a lawful noncitizen appear on any of them?
• What year and denominator accompany the headline number?
• Did an election agency audit establish unlawful registration or voting, or did a post infer it?