美国中期选举收官阶段之际，特朗普政府已向多个州提交所谓“选民登记名单中存在非公民”的名单，加剧其旨在阻止这些人在选举中投票的争议性努力。
在加利福尼亚州，政府声称登记名单中可能有2万人是非公民；而在华盛顿州，政府本周指控，400名潜在非公民选民在过去选举中累计投票1500次。司法部向两个州发信，要求其扣押司法部所称的可能为非公民的人的选票——但加州称该要求违反其法律，司法部遂起诉加州。
本月对加州和华盛顿的要求，标志着政府通过创建潜在非公民选民名单来追查选民舞弊的行动显著升级，这也是联邦政府试图证明唐纳德·特朗普总统所谓“大规模选民舞弊”的毫无根据的说法的众多方式之一。
这些指控引发了关于政府如何创建潜在非公民选民名单的质疑，同时也招致批评：在审查之下，其关于数万名非公民选民的说法已大幅缩水。
州选举官员和选举专家对特朗普政府用来作出这些全面指控的数据的有效性提出质疑，也对在中期选举提前投票已经开始之际向各州提出这些要求的决定提出质疑。
周二加州案件的地区法院听证会上，州务卿办公室、洛杉矶县及外部团体的律师辩称，联邦政府的数据已经过时、存在缺陷，是“剥夺美国公民选举权的隐患”。
“这是一个令人深感不安的程序，试图将社会保障号码与选民档案绑定，以‘钓鱼’式排查任何拥有外国人登记号码的人，”代表拉丁裔美国公民联盟和全国有色人种协进会发言的律师安娜·鲍德温说。
加州中区联邦检察官办公室首席助理检察官比尔·埃塞利辩称，联邦政府已掌握关于登记名单中非公民的“可信证据”，而加州拒绝处理。
埃塞利称，联邦政府试图“保护合法公民的选票”。
尽管案件的核心是，在宪法主要将选举管理权赋予各州的情况下，联邦政府能否优先于州选举法，但政府编制选民名单的过程也经常成为争议焦点。
埃塞利提议，提交宣誓书的国土安全部官员可在周二听证会上作证，以帮助解释用于确定名单上选民非公民身份的程序。
但由克林顿政府任命的美国地区法官戴维·卡特表示，他将指定一名特别主事官，在加州12月3日纠正选票的最后期限之前，调查联邦政府和洛杉矶县提供的数据。
在周二听证会之前，第九巡回上诉法院一个由三名法官组成的合议庭于周一以2比1的裁决，以程序为由驳回了司法部对加州的诉讼。但两名支持加州的法官——特朗普任命的马克·班尼特和拜登任命的加布里埃尔·桑切斯——都在异议意见中写道，他们也会作出不利于司法部的裁决。
桑切斯在意见书中写道，加州和洛杉矶县“对美国数据和生成该名单的方法的准确性和可靠性表达了严重疑虑，并认为这一非同寻常的要求可能会剥夺数千名加州选民的选举权”。
他在周日的紧急听证会上针对司法部用来提出指控的数据和名单，以及特朗普政府如何得出这些结论提出了几个尖锐问题。
“你的说法很大程度上取决于该名单的可靠性，”桑切斯对司法部律师马修·唐纳利说。
特朗普政府称，各州选民登记名单中的非公民数量更大。在特朗普7月关于选举舞弊的黄金时段演讲后，国土安全部部长马克韦恩·马伦声称，加州有超过19万名登记选民“可能”是非公民。
加州是马伦在特朗普演讲后点名的几个民主党执政州之一。
“我们已在四个州确认了25万名登记投票的非公民，”马伦对记者说，所指的是加州、新泽西州、宾夕法尼亚州和内华达州的调查结果。
尽管马伦将这些数字表述为事实，但数月来，这些数字在审查下已大幅缩水。
例如，在内华达州，国土安全部声称在该州选民登记名单中确认了15903名非公民。但在后续会议上，国土安全部官员告诉该州，通过人工复核，他们“确认”内华达州仅有185名登记者是非公民。
内华达州选举官员反驳称。“至于这185人，我们的记录显示他们是公民；你没有提供任何实质证据证明并非如此，”一名内华达州官员上月致信国土安全部。
不过，司法部律师唐纳利在周日听证会上对此提出异议，称内华达州领导人的说法仅基于选民登记时宣誓为公民的声明。
“坦白说，我们认为这并不构成我们的名单无效的证据，”他说。
当被问及加州的相关数据从19万选民降至目前所称的2万时，唐纳利在周日听证会上表示，这一下调数字是国土安全部进一步核查的结果。
在对民主党和共和党执政州的1亿多份选民记录进行审查后，国土安全部仍声称已发现30万名“潜在”非公民。截至9月中旬，该机构已根据审查展开1620项调查，迄今已导致约160人被捕。
在周二的法庭听证会上，卡特对埃塞利表示，政府已有处理投票的非公民的办法。“看在上帝的份上，你有补救措施，去提起公诉吧，”法官难以置信地说。
国土安全部一名发言人在一份声明中表示，“19万仍然是国土安全部要求（州务卿雪莉·）韦伯审查的加州潜在非公民登记选民数量”。
“司法部提到的2万人是迄今经过人工复核的登记者子集，”该发言人说，并补充称，国土安全部的这一程序是“保障我们选举的明智步骤”。
国土安全部在给各州的信函中未说明如何得出其认为是非公民的选民名单细节，但更多细节已在加州诉讼中披露。
美国公民及移民服务局和国土安全调查局官员在法庭文件中表示，他们于2025年8月获取了加州的公开选民名单。虽然该选民名单来自州务卿办公室，但司法部未披露具体是谁从该州获取了名单。
特朗普政府“获得了一份从加州购买的名单”，唐纳利在周日听证会上说。
今年早些时候，国土安全部开始分析该文件，使用第三方系统进一步分析数据，并与国土安全部移民数据库比对，以识别潜在非公民。随后，国土安全部官员称，他们将核查结果与联邦数据库比对，并对选民档案进行人工复核。
“这些程序最终形成了一份缩减后的名单，上面是可能成为非公民并计划参加2026年联邦选举投票的人，”美国公民及移民服务局欺诈检测和国家安全局副局长尼古拉斯·巴特尔写道。
两党政策中心选举项目常务董事雷恩·奥雷表示，第三方信息似乎被用于“基于公开可用的选民名单信息补充个人可识别信息”。
“在我看来，这似乎存在很高的误报匹配风险，”奥雷说。
尽管唐纳利在周日的第九巡回上诉法院听证会上表示，他不知道使用了哪些第三方，但巴特尔本周在宣誓书中写道，美国公民及移民服务局通过商业数据公司律商联讯的Accurint系统“增强”了选民登记名单数据。
加州民主党参议员亚历克斯·帕迪拉收到的一份whistleblower报告中，whistleblower称，源数据中未包含的某些信息是通过一个名为“补充魔法”的流程纳入的。
司法部依赖的是加州公开可用的选民名单，因为该州拒绝了司法部获取其未删节选民文件的请求，而未删节文件将包含社会保障号码等敏感信息。司法部已寻求获取几乎每个州的未删节选民名单，并起诉了约30名拒绝移交的州，但迄今在所有相关法院案件中均败诉。
州政府律师克里斯蒂娜·麦卡尔辩称，政府的解释未能充分说明政府如何编制数据，甚至未说明如何获取该州的选民登记名单。
“它们不可靠、模糊，且缺乏个人知情依据，”她在谈到政府指控的非公民选民名单时说。
巴特尔在宣誓书中称，美国公民及移民服务局将其编制的数据与国务院和社会保障管理局的数据库比对，以剔除名单上拥有护照、可确认其公民身份或已死亡的美国公民。
移民机构随后为每个潜在非公民匹配结果分配“置信水平”，再进行人工复核。这些案件随后被移交国土安全调查局作进一步调查。
在另一项诉讼中，联邦法院上周下令国土安全部停止其追查非公民选民的某些调查行动；司法部律师凯尔西·赫兰辩称，提起诉讼的团体没有证据表明国土安全部从疑似非公民选民大名单缩小至被调查的小名单的过程中影响了美国公民。
“这是初步步骤，经过数据库交叉比对等多种来源核查，美国公民及移民服务局官员也有保障措施，即若某人看似是公民则不应继续推进，”赫兰上周在地区法院听证会上说。
在该诉讼中，一份未完全公开的国土安全调查局官员文件提到了该机构数据可能存在的不准确之处。“当美国公民及移民服务局名单中发现潜在不准确时，国土安全部寻求进一步资源以防止不准确，并开发了生成国土安全调查局非法投票调查线索的新模型，”该官员写道。
过去一年，国土安全部还试图通过其“外国人身份核对系统”（SAVE）识别选民登记名单中的非公民，尤其是在自愿移交未删节选民文件的州。最高法院上月允许联邦政府继续使用SAVE进行调查，但国土安全部表示，该数据库未用于其发送给加州的名单编制。
选举专家表示，目前仍不清楚国土安全部的匹配结果可靠性如何，特别是考虑到特朗普政府使用的是公开可用的选民文件，而非包含社会保障号码等个人识别信息的州名单。
“我们仍然不知道使用了哪些数据库，更重要的是，匹配到何种程度才算充分，”圣母大学法学院选举法教授德里克·马勒说。
加州在法庭文件中辩称，用于编制其认为可能是非公民的2万名选民名单的国土安全部选民登记名单基于一年多前的旧数据，可能存在误报。
“政府自身的宣誓书描述了一项大量数据匹配工作，始于一年多前、来自未具名来源的选民登记名单，补充了未具名商业平台的信息，并与联邦数据库比对，”洛杉矶县登记主任迪恩·洛根在该案的事实宣誓书中写道。
“其他州的经验证实了为什么这种匹配需要调查和基于个人的证明，”洛根补充道。
https://www.cnn.com/
洛根的办公室从洛杉矶县地址所属的7035个名字中抽取了200名选民样本。负责IT和技术的助理登记/记录员阿曼·布拉尔在宣誓书中写道，其办公室发现，在这200人中，42.9%已取消选民登记，4.6%被列为不活跃，意味着今年不会向其邮寄选票。
布拉尔表示，超过71%的样本从未投过票，在有出生地信息的人中，38.9%申报出生于美国。
华盛顿州已收到司法部周二发出的信函，该州州务卿办公室指出，400条记录中有一条来自1994年。“司法部使用的系统经常将合法美国公民标记为非公民，危及他们的法定投票权，”州务卿办公室在一份声明中说。
在本周提交的一份新法庭文件中，国土安全部官员反驳了加州抽取的200名选民样本。巴特尔在宣誓书中写道，美国公民及移民服务局审查了国土安全部名单上的51名选民，“其中加州的选民登记名单显示该人声称出生于美国”。
“经全面核查县样本名单上所有个人的移民记录，根据美国公民及移民服务局的记录，确认美国公民及移民服务局匹配到县样本名单的个人均未出生于美国，”巴特尔写道，不过他承认，其中4名选民“可能有权获得美国公民身份”，尽管他们并非出生在美国。
CNN记者蒂尔尼·斯尼尔对本文报道有贡献。
In the closing days of the midterms, the Trump administration has sent multiple states lists of what it alleges are noncitizens on the voter rolls, intensifying its controversial efforts to stop those people from voting in the election.
In California, the administration claimed there were 20,000 people who could be noncitizens on the rolls, while in Washington state, the administration alleged this week that 400 possible noncitizen voters had voted 1,500 times in past elections. The Justice Department sent letters to both states demanding they set aside the ballots of those the department claimed could be noncitizens — and sued California when the state said that the request ran afoul of its laws.
The demands this month of California and Washington marked a significant escalation of the administration’s efforts to hunt voter fraud by creating lists of potential noncitizen voters, one of numerous ways the federal government has tried to prove President Donald Trump’s baseless claims of mass voter fraud.
The allegations have sparked questions about how the administration created its lists of voters who could be noncitizens — and criticism that its claims of tens of thousands of noncitizens voters have whittled down under scrutiny.
State election officials and election experts have questioned the validity of the data the Trump administration is relying on to make its sweeping allegations — in addition to the decision to bring these demands to the states as early voting in the midterm election is already underway.
At a district court hearing Tuesday in the California case, attorneys for the secretary of state, Los Angeles County and outside groups argued that the federal government’s data was outdated, flawed and was a “recipe for disenfranchisement” of US citizens.
“This is a deeply troubling process, that is trying to use social security numbers that they’ve glued to the voter file, to go fishing for anyone who has an alien number,” said Anna Baldwin, an attorney arguing on behalf of the League of United Latin American Citizens and NAACP.
Bill Essayli, first assistant US attorney for the Central District of California, argued that the federal government had “developed credible evidence” of noncitizens on the rolls that California refused to address.
Essayli argued that the federal government was trying to “protect votes of lawful citizens.”
While much of the case focused on whether the federal government could preempt state election law when the Constitution primarily gives the power to administer elections to the states, the circumstances surrounding how the government came up with the lists of voters were also a frequent point of contention.
Essayli offered to have the DHS officials who submitted sworn statements testify at Tuesday’s hearing to help explain the process that was used to determine the voters on the list were not citizens.
Instead, US District Judge David Carter, a Clinton appointee, said he’s appointing a special master to dig into the data provided by both the federal government and Los Angeles County, ahead of the state’s December 3 deadline to cure ballots.
Ahead of Tuesday’s hearing, a three-judge panel of the 9th Circuit US Court of Appeals rejected the Justice Department’s California lawsuit on Monday in a 2-1 decision on procedural grounds. But the two judges who sided with the state, Trump appointee Mark Bennett and Biden appointee Gabriel Sanchez, both wrote in concurring opinions they also would have ruled against DOJ.
In his opinion, Sanchez wrote that California and LA County “have expressed serious doubts about the accuracy and reliability of the United States’ data and methodology used to generate this list and contend that this extraordinary request threatens to disenfranchise thousands of California voters.”
He asked several pointed questions during Sunday’s emergency hearing about the data and list the Justice Department was using to make its claims and how the Trump administration arrived at its conclusions.
“Your claims very much depend on the reliability of that list,” Sanchez said to DOJ attorney Matthew Donnelly.
The Trump administration has claimed even larger numbers of noncitizens on state voter rolls. After Trump’s July primetime speech on election fraud, Department of Homeland Security Secretary Markwayne Mullin claimed there were more than 190,000 voters registered in California who “may be” noncitizens.
California was one of several Democratic-run states that Mullin singled out after Trump’s speech.
“We’ve identified 250,000 noncitizens registered to vote in just four states,” Mullin told reporters, citing findings in California, New Jersey, Pennsylvania and Nevada.
While Mullin stated the figures as facts, those numbers have shrunk considerably under scrutiny in the months since.
In Nevada, for instance, DHS alleged it had identified 15,903 noncitizens on the state’s voter rolls. But in a follow-up meeting, DHS officials told the state they had “confirmed” just 185 Nevada registrants as noncitizens through manual review.
Nevada election officials pushed back. “As for the 185 individuals, our records indicate that they were citizens. You have provided nothing of substance to suggest otherwise,” a Nevada official wrote to DHS last month.
Donnelly, the DOJ attorney, disputed this at Sunday’s hearing, however, saying that Nevada’s leaders were basing this solely on voters attesting they were citizens when they registered.
“That, candidly, we do not think constitutes proof that our list was invalid,” he said.
Asked about the downward shift in California from 190,000 voters to the 20,000 now being alleged, Donnelly said during Sunday’s hearing that the lower number was the result of further checks by DHS.
After a review of more than 100 million voter records across both Democratic- and Republican led states, DHS is still claiming it has found 300,000 “potential” noncitizens. As of mid-September, the agency had opened 1,620 investigations based on the review, and so far, the effort has led to around 160 arrests.
During Tuesday’s court hearing, Carter told Essayli that the government already had a way to handle noncitizens who voted. “You’ve got a remedy for goodness sakes, go out and indict,” the judge said incredulously.
A DHS spokesperson said in a statement that “190,000 remains the number of potential noncitizens registered to vote in California that DHS asked Secretary (of State Shirley) Weber to review.”
“The 20,000 referenced by DOJ is a subset of registrants that have gone through the manual verification process so far,” the spokesperson said, adding that the DHS process was a “commonsense step to secure our elections.”
DHS has been vague in correspondence to states about how it came up with its lists of voters it believes are noncitizens, but more details have emerged in the California lawsuit.
USCIS and Homeland Security investigations officials have said in court filings they obtained California’s public voter list in August 2025. While the voter list was obtained from the secretary of state’s office, DOJ redacted who specifically obtained it from the state.
The Trump administration “was supplied a list that was purchased” from California, Donelly said during Sunday’s hearing.
Earlier this year, DHS began analyzing the file, and it used a third-party system to further analyze the data and compare it to DHS immigration databases to identify potential noncitizens. Then, DHS officials say they checked the findings against federal databases and conducted manual checks of voter files.
“These processes resulted here in a culled list of people who are likely to be noncitizens and who registered to vote in the upcoming 2026 federal election,” wrote Nicolas Bartell, associated director of Fraud Detection and National Security Directorate at USCIS.
Wren Orey, managing director of the Bipartisan Policy Center’s Elections Project, said the third-party information appeared to be used to “backfill personally identifiable information based on the information on the publicly available voter list.”
“That seems to me like a big risk that they were getting high rates of false positive matches,” Orey said.
While Donnlly said he did not know the third parties used during Sunday’s 9th Circuit hearing, Bartell wrote in his declaration this week that USCIS “enhances” the voter roll data through Lexis Nexis’ Accurint, a commercial data company.
In a whistleblower report filed with Democratic Sen. Alex Padilla of California, the whistleblower alleged that certain information not included in the source data was brought in through a process referred to as “supplemental magic.”
DOJ is relying on a voter list in California that’s publicly available, because the state rejected the Justice Department’s attempt to obtain its unredacted voter file, which would contain sensitive information like Social Security numbers. DOJ has sought unredacted voter lists from nearly every state and has sued roughly 30 states that refused to turn them over but has lost every court decision on the matter so far.
Christina McCall, an attorney for the state, argued that the government’s explanation doesn’t sufficiently spell out the way that the administration compiled its data — or how it even obtained the state’s voter rolls.
“They’re not reliable, they’re vague, and they lack personal knowledge,” she said of the state’s lists of alleged noncitizen voters.
Bartell stated in his declaration that USCIS compares the data it compiles against databases from the State Department and Social Security Administration to weed out US citizens on its lists who have a passport, confirming they’re citizens, or have died.
The immigration agency then assigns a “confidence level” for each possible noncitizen match, followed by a manual review. Those cases are then referred to HSI for further investigation.
In separate litigation, which led to a federal court last week ordering DHS to stop certain investigative actions in its hunt for noncitizen voters, DOJ attorney Kelsey Helland argued that the groups suing did not have evidence that US citizens were impacted by the steps DHS took going from large list of suspected noncitizen voters to a smaller list of individuals who are investigated.
“That’s the initial step, and it goes through various sources of cross-checking databases, and there’s safeguards where USCIS officers are instructed not to keep going if it appears that a person is a citizen,” Helland said during a District Court hearing last week.
In that litigation, a redacted HSI official touched on the potential inaccuracies in the agency’s data. “When potential inaccuracies were identified in the USCIS lists, DHS sought further resources to prevent inaccuracies and developed a new model for generating leads for HSI’s unlawful voting investigations,” the official wrote.
Over the past year, DHS has also sought to identify noncitizens on voter rolls through its Systematic Alien Verification for Entitlements database — known as “SAVE” — particularly in states that voluntarily turned over their unredacted voter files. The Supreme Court allowed the federal government to continue using SAVE for its investigations last month, but DHS said that the database was not used to compile the list it sent California.
Election experts say that it’s still not clear how reliable the DHS matches may be, especially given that the Trump administration is using a publicly available voter file, not the state’s list that contains personally identifying information like Social Security numbers.
“We still don’t have all the information about what kinds of databases are used. And then more to the point, what level of matching is sufficient,” said Derek Muller, an election law professor at Notre Dame Law School.
California has argued in court filings that the DHS voter file used to compile the list of 20,000 voters it believes could be noncitizens is based on data more than a year old and could be subject to false positives.
“The Executive’s own declarations describe a bulk data match that began with a voter roster more than a year old, obtained from an unidentified source, supplemented through an unnamed commercial platform, and compared against federal databases,” Los Angeles County Registrar Dean Logan wrote in a factual declaration for the case.
“Other states’ experience confirms why such matches require investigation and individualized proof,” Logan added.
https://www.cnn.com/
Logan’s office took a sample of 200 voters from the 7,035 names with LA County addresses. His office found that of the 200 individuals, 42.9% had canceled voter registration records and 4.6% were considered inactive, meaning they were not mailed ballots this year, Aman Bhullar, an assistant registrar/recorder who works on IT and technology, wrote in a declaration.
Bhullar said that more than 71% of the sample had never voted, and 38.9% of those who had a birthplace reported indicated they were born in the US.
In Washington state, which received the letter from the Justice Department Tuesday, the secretary of state’s office noted that one of the 400 records was from 1994. “The system the DOJ uses frequently flags legitimate U.S. citizens as noncitizens, jeopardizing their legal right to vote,” the secretary of state’s office said in a statement.
In a new court filing this week, DHS officials pushed back on California’s sample of 200 voters. Bartell wrote in a declaration that UCSIS reviewed 51 voters on the DHS list “where California’s voter roll indicates the person claimed to be born in the United States.”
“After a thorough review of the immigration records of all the individuals on the County’s Sample List, it was determined that none of the individuals USCIS matched to the County’s Sample List were born in the United States according to USCIS records,” Bartell wrote, though he acknowledged four of the voters “might have a claim to U.S. citizenship,” even though they weren’t born in the United States.
CNN’s Tierney Sneed contributed to this report.