US News Bulletin report

A Judge Stopped a Voter Investigation Built on Bad Lists


The people being investigated included citizens who had every legal right to vote. That is the part that stays with me.

A federal judge in California has ordered the Trump administration to halt investigations in nine states into whether noncitizens are registered to vote. U.S. District Court Judge Araceli Martinez-Olguin, sitting in Oakland, found that the government relied on data full of errors. She also found that the effort was intimidating groups that help people vote.

The order covers California, Connecticut, Georgia, Missouri, Nevada, New York, Pennsylvania, Washington, and Wisconsin. It is set to last until polls close on Election Day, November 3. The judge briefly delayed the order until Friday evening so the administration could appeal.

What the judge found

The investigation began with lists of people the government believed might have registered or voted before becoming citizens. Those lists came from state and local records, public information, and commercial voter-roll data. The judge called them voluminous and unreliable.

Her concern was not abstract. There is no single federal database that holds the citizenship status of everyone living in the United States. So a record can look suspicious and still be wrong. A person may have naturalized years ago. A name may match another person. A date may be incomplete or entered badly.

During what court records described as a surge, officers were limited to about 12 minutes to review each person. The judge wrote that the time limit made the risk of error worse. That detail is hard to brush aside. Twelve minutes is not much time to untangle a person’s citizenship history from imperfect records.

The order does not say the government must stop every inquiry into illegal voting. It blocks certain election-related field operations and interviews in the nine states when agents know, suspect, or reasonably should suspect that the person is a U.S. citizen. It also blocks operations based wholly on the government’s voter-matching records when the target is known or reasonably suspected to be a citizen.

That distinction matters. A government can investigate suspected fraud. The question in this case was whether it could do so with lists the judge found flawed, in the final days before an election, in ways that frightened people who were trying to help others vote.

Why voter helpers were afraid

The lawsuit was brought by the California Nurses Association and National Nurses United, along with the Voto Latino Foundation, San Francisco, and Santa Clara County. They argued that the federal actions were interfering with their members’ work encouraging people to take part in the 2026 midterm elections.

The judge agreed that the unions had shown a likelihood of success on their claim that the policy intimidated people who assist or encourage others to vote. She found that the use of unreliable data, combined with highly publicized arrests and interactions with armed officers, could create a reasonable fear of law enforcement trouble.

That fear does not have to be universal to cause damage. If a nurse, community volunteer, or organizer wonders whether helping a neighbor register could bring an agent to the door, the practical effect is the same. Some people will step back. Some will decide the work is not worth the risk.

The Voting Rights Act bars intimidation, threats, or coercion against people who assist voters or take part in voter mobilization. The judge found the government’s approach likely violated that protection.

What remains unsettled

It is worth being careful about what this ruling does and does not decide.

The judge did not make findings about the extent of the threat of voter fraud. She noted that other courts have found the federal government has failed to show that significant fraud threatens the reliability of upcoming elections. But this order was not a final verdict on every claim in the case.

The Department of Homeland Security has said it is focused on removing noncitizens who vote and dilute the votes of U.S. citizens. It also said it had no plans to question or investigate people it believes are U.S. citizens. That statement and the judge’s findings point in different directions, and the appeal process may test which view controls.

What is documented is the order itself: the judge found error-prone data and intimidation risks, and she restricted specified field operations in nine states through Election Day. What remains disputed is how accurate the underlying lists were, how widespread the intimidation was, and whether the administration’s methods can survive legal review.

Why this should worry all of us

I understand the impulse behind voter-fraud investigations. Elections only work if people believe the rules are real and evenly applied. But a belief in clean elections cannot become a reason to use sloppy lists against people who may be citizens.

The danger is not only that someone could be questioned by mistake. It is that the mistake arrives with force: an agent at the door, a public arrest video, a story that spreads faster than any correction. For a person with limited time, money, or English, even a wrongful encounter can feel like a warning.

And the people most likely to feel that warning are not necessarily the people the government intended to reach. They are the ones nearby: family members, coworkers, neighbors, and the volunteers who show up to make voting easier.

A fair election needs two things at once. It needs rules against fraud. It also needs room for citizens to encourage one another to vote without wondering whether a knock at the door follows. The judge’s order does not settle every argument about noncitizen voting. It does say that, in the days before an election, the government cannot build an enforcement campaign on records it knows are shaky and then call the fear it produces incidental.

More on: Politics

That seems like a sensible place to draw the line.