The Texas secretary of state’s office on Friday called into question the citizenship status of 95,000 registered voters who were found to have identified themselves at some point to a state law enforcement agency as noncitizen, legal residents of the United States.
The office said its findings were a result of an 11-month investigation with the Texas Department of Public Safety that also found that about 58,000 people on the list had voted since 1996. The results of the investigation were referred on Friday to Attorney General Ken Paxton, who said he planned to open a potentially sprawling investigation.
The two announcements seemed certain to reignite partisan debates over the frequency and impact of voter fraud, which Republicans have claimed is rampant in America. Democrats scoff at that notion, and a voter fraud commission started by (and later angrily disbanded by) President Trump found no evidence of widespread electoral fraud.
“Every single instance of illegal voting threatens democracy in our state and deprives individual Texans of their voice,” Mr. Paxton, a firebrand conservative who has prosecuted isolated cases of illegal voting with gusto, said in a statement. “Nothing is more vital to preserving our Constitution than the integrity of our voting process, and my office will do everything within its abilities to solidify trust in every election in the state of Texas.”
But Democrats and voting rights advocates were skeptical of the state’s claims. More than 8.3 million people voted in the Texas governor’s race last year, which means that even if all 58,000 people who voted were, in fact, found to be noncitizens and voted in 2018 — a claim that no state official has made — they would have amounted to only 0.69 percent of all votes that were cast.
“Because we have consistently seen Texas politicians conjure the specter of voter fraud as pretext to suppress legitimate votes, we are naturally skeptical,” Representative Rafael Anchia, a Democrat in the Texas House of Representatives, said in a statement.
Kristen Clarke, the president of the Lawyers’ Committee for Civil Rights Under Law, also cautioned that the state’s findings, and the speed with which the attorney general raised the specter of prosecution, could foreshadow an attempt at voter suppression.
“Texas has a rich history of undertaking action to make it harder for people to vote,” she said. “Whenever you’re invoking the threat of criminal prosecution, the chilling effect becomes almost unavoidable.”
Her organization sued Texas over its 2017 voter registration law, which a federal appeals court ruled last year was not discriminatory against black and Latino voters. She described the law on Friday as “one of the most restrictive voter ID laws in the nation.”
Sam Taylor, a spokesman for the secretary of state, said the announcement on Friday did not mean that the authorities had discovered 95,000 registered voters who it knew for a fact were noncitizens.
Instead, the office was advising local officials to ask these voters via mail to provide proof of their citizenship. Mr. Taylor noted that it is a felony for noncitizens to vote in Texas, and a misdemeanor for them to register.
Mr. Taylor said the authorities believed that the method they had used to identify these potentially problematic voter registrations had left “very little, if any” room for error.
He said the investigation used a range of identifiers — including first and last names, birth dates and full or partial Social Security numbers — to cross-reference voter rolls and public safety records.
For example, he said, if someone had used a green card to apply for a driver’s license, that would remain on file with the Department of Public Safety, which is responsible for law enforcement and vehicle registration.
“We can’t see a situation in which this would produce a false positive,” he said. “These are people whose last and most recent visits to D.P.S. showed them to be noncitizens through documentation that they submitted and which D.P.S. has kept on file.”
But Ms. Clarke said there were often mundane explanations for cases like these, including simple administrative error.
“In our experience, state databases are often riddled with errors, are not up-to-date and don’t reflect the most recent information with regard to individuals,” she said. “It will be very important for the state to provide more information about how they carried out their analysis, and we will have to test the veracity of their claims.”