Republican Sen. Mike Lee said that he believes there are “at least tens of thousands, probably hundreds of thousands” of noncitizens illegally registered to vote in the U.S., adding that a federal tool used in nearly two dozen states would help identify the number. But the tool has wrongly flagged many as being noncitizens, and there’s no evidence of widespread noncitizen voting.
The data-matching program employed in those states over the last year identified about 10,000 potential noncitizens on voter registration lists, out of about 49 million voter registrations checked, according to reporting by the New York Times citing federal officials. But upon further investigation, county officials found U.S. citizens were among those identified.
In addition, election officials determined some noncitizens were inadvertently added by county officials to voter lists, and still others were noncitizens who mistakenly checked a box for voter registration even after acknowledging on the same forms that they were noncitizens.
Experts and state audits refute the idea of widespread noncitizen voting.
The SAVE America Act championed by Lee — and touted by President Donald Trump as necessary to stop illegal voting by noncitizens — would require all states to submit their voter registration lists to the Department of Homeland Security to be run through this tool, called the Systematic Alien Verification for Entitlements, or SAVE, program. The bill passed the House and is being debated in the Senate. Lee made his comments about noncitizens being registered to vote in a March 22 interview on Fox News’ “Sunday Morning Futures.”
Jasleen Singh, a senior counsel and manager in the Brennan Center for Justice’s democracy program, said that Lee’s speculation about the number of noncitizens on voter registration lists amounts to “another outlandish claim without evidence.” The reality, Singh said, is that “noncitizen voting is vanishingly rare.”
The SAVE program, Singh said, “is one of many tools that election officials have in their toolbox to use. It comes with a myriad of data flaws, and any results that come directly from a search to the SAVE program need to be viewed with that lens and with a good degree of skepticism.”
Acting upon an executive order from Trump in March 2025, DHS overhauled the SAVE program last spring to include Social Security data. Trump also waived fees to states to access the database, allowing bulk searches.
“What we do know is that in states that have started reviewing the voter registration files in order to weed out those [ineligible people] who might have registered, perhaps inadvertently … already there have been thousands of voter registration files identified in just the handful of states doing their own reviews,” Lee told the Hill on March 20.
We reached out to Lee’s office but did not get a response.
Many states — predominantly ones run by Democrats — have refused to share their voting lists with the SAVE program. U.S. Attorney General Pam Bondi has sued 29 states and the District of Columbia for failing to provide the federal government with their lists.
But nearly two dozen states have utilized the SAVE program. Lee is correct that “thousands” of people have been flagged as potentially being noncitizens. As we said, of the 49.5 million voter registrations checked, DHS referred about 10,000 cases to investigators, according to a Jan. 14 New York Times report that attributed the figures to a spokesman for U.S. Citizenship and Immigration Services. (There were 174 million people registered to vote in the U.S. for the 2024 election, according to the U.S. Census. In other words, less than a third of all names on state voter registration lists nationwide have been run through the SAVE program.)
But the Times reported that local election officials began to discover that some of the names flagged by the SAVE program turned out to be citizens. That appeared to be particularly true for recently naturalized citizens. Tens of thousands of people are naturalized as citizens every month, according to U.S. Citizenship and Immigration Services data.
A joint investigation by ProPublica and the Texas Tribune found that in addition to many citizens being wrongly flagged as noncitizens, several election officials “came across instances in which voters marked on registration forms that they weren’t citizens, but were registered by election office staffers in error. Clerks also said voters have told them they’d misunderstood questions about eligibility when getting drivers’ licenses.”
Ongoing research by the Center for Election Innovation & Research “continues to find that sweeping allegations about noncitizen registrations or voting appear to arise from misunderstandings, mischaracterizations, or outright fabrications about complex voter data. In every examined case, when claims about large numbers of noncitizens on voting rolls are subject to scrutiny and properly investigated, the number of alleged instances falls drastically.”
Even in the states that have used the federal SAVE program, “Claims of large numbers of possible noncitizens on voter records are revised significantly downward after proper investigation and scrutiny. Most often, investigations into large claims reveal that at least some early flags were based on outdated, incomplete, or improperly matched data that incorrectly labeled eligible citizens as possible noncitizens,” CEIR reported in February. Those smaller, revised numbers “generally receive far less public attention.”
Interestingly, the SAVE America Act faces significant opposition from the top Republican election official in Lee’s home state of Utah, which last year initiated a citizenship review of all registered voters in the state. Ultimately, officials announced in January that they were only able to confirm the state voter rolls included one noncitizen, and that person did not vote.
State officials first compared voter records against driver’s license data, which records citizenship status. The conclusion: 99.9% of the state’s 2 million voters were citizens. But that left the status of 71,314 people unclear, so officials checked those against the SAVE database, which narrowed the potential number of noncitizens to 8,836. Staff in the state elections office then reviewed the remaining voters’ information. That narrowed the list to 486 they could not immediately verify were citizens. Officials sent letters to everyone in that group and got back 52 responses, including many from older voters who registered before the state required a driver’s license or Social Security number.
“The bottom line is, there is not a widespread problem,” Lt. Gov. Deidre Henderson, a Republican, said at the time. “You hear people say hundreds or thousands — it’s just not.”
Henderson, who oversees elections in the state, wrote in a January press release that through Utah’s citizenship review, “We also learned that the federal government does not keep accurate databases.”
The SAVE program, she said, “is notoriously inaccurate and frequently flags individuals who are, in fact, citizens.”
Henderson and others have also raised concerns about the SAVE America Act requiring states to use the DHS database immediately, in the midst of a midterm election year.
“If we want a federal law mandating voter ID or DPOC [documentary proof of citizenship], and it’s really not about disenfranchising a bunch of voters, then states and voters need an onramp with time to prepare — get the documents, obtain the right ID, set up the system,” Henderson wrote in a social media post on March 17. “That’s not what’s happening with the SAVE America Act. This bill would be effective immediately in the middle of an election year.”
Similar stories have played out in other states that used the SAVE program.
One of the first states to implement the SAVE program was Texas, and on Oct. 22, Texas’ secretary of state, Jane Nelson, announced that it had completed a full comparison of the state’s voter registration list against citizenship data in the SAVE database. Calling it a “game changer,” Nelson said the SAVE program identified 2,724 potential noncitizens on the state’s voter registration rolls — or less than 0.02% of more than 18 million voters.
Nelson said the list of those potential noncitizens was sent to Texas counties to conduct investigations, with the understanding that those deemed to be noncitizens would be purged from voter registration lists and those who were found to have voted illegally would be referred to the Texas attorney general for prosecution.
“Everyone’s right to vote is sacred and must be protected. We encourage counties to conduct rigorous investigations to determine if any voter is ineligible — just as they do with any other data set we provide,” Nelson said.
But that’s where things began to fall apart.
As a joint investigation by ProPublica and the Texas Tribune documented, lacking clear guidance, some counties investigated; others didn’t. Some sent letters to people on the list and purged those who failed to respond; others didn’t purge any names.
Some counties compared the names on their list to databases kept by the Department of Public Safety, which requires proof of citizenship if residents register to vote when obtaining a driver’s license. Those checks found many of those on the list identified as potentially noncitizens were citizens.
In Potter County, for example, three of nine voters on the list had proof of citizenship on file, the ProPublica/Texas Tribune investigation found. In Travis County, it was 11 of the 97 voters flagged by the SAVE program. Overall, the counties that checked the SAVE-generated list against DPS records found “more than 5% of the voters SAVE identified as noncitizens proved to be citizens,” the investigation concluded.
“It has proven to be inaccurate,” Travis County’s voter registrar, Celia Israel, told the publications. “Why would I rely on it?”
While the SAVE program accurately identified many on the voter registration rolls who were ineligible to vote, “Several [counties] came across instances in which voters marked on registration forms that they weren’t citizens, but were registered by election office staffers in error. Clerks also said voters have told them they’d misunderstood questions about eligibility when getting drivers’ licenses,” the ProPublica/Texas Tribune report said.
In Louisiana, the SAVE program identified 403 potential noncitizens registered to vote, out of 2.96 million registered voters. That’s about 0.014%. Of those potential noncitizens, 83 cast at least one vote going back to the 1980s (though it was not clear how many of those were later verified to be noncitizens). In 2024, 2,006,975 people voted in the presidential election in Louisiana. Even if all 83 of them voted that year, that would translate to about 0.004% of all votes cast in the state.
“I want to be clear: noncitizens illegally registering or voting is not a systemic problem in Louisiana,” Louisiana Secretary of State Nancy Landry said when the preliminary results were revealed last September.
Missouri also employed the SAVE program and generated lists of potential noncitizens, which it then circulated to local officials.
On Dec. 3, more than 70 county election clerks from both parties wrote a letter to the state’s speaker of the House warning, “These lists are deeply flawed: they are outdated, inaccurate, and include individuals we know to be U.S. citizens—our neighbors, colleagues, and even voters we have personally registered at naturalization ceremonies.”
It’s not clear how many noncitizens flagged by the SAVE database actually voted. But there have been relatively few arrests nationwide for illegal voting by noncitizens.
That makes sense, Singh told us, considering the stiff consequences for convictions for voting illegally as a noncitizen. Current federal law requires those registering to vote to attest that they are citizens under penalty of perjury. Noncitizens convicted of voting in federal elections face fines, jail time and deportation.
“Someone who is in this country, who may not have documents, or who has a legal presence and is not a citizen yet, whatever it is, they’re not going to risk their ability to be in this country to cast a ballot, because they will be subject to deportation,” Singh said. “And it’s just not a risk that folks are, if we think about it logically and reasonably, that folks are going to be willing to take.”
According to the conservative Heritage Foundation’s election fraud database, just under 100 noncitizens have been convicted of illegally voting or registering to vote since 1982.
There may be so few prosecutions, Singh said, because by and large, when noncitizens are on registration rolls “it’s likely a mistake or because of an error by the person registering, or maybe the DMV … whatever it is, it’s a mistake rather than an actual intentional act.”
“The evidence is that the number of noncitizens illegally voting in federal elections is extremely low, not high enough to have changed the party outcome of any federal election in recent years,” Walter Olson, a senior fellow at the libertarian Cato Institute, told us last April. “Audits and investigations in states like Ohio, Nevada, and North Carolina have found the numbers to be tiny in relation to votes cast. … The consistent experience has been that very few persons in this category mistakenly or deliberately vote.”
Senate Minority Leader Chuck Schumer has said the SAVE America Act “could disenfranchise over 20 million American citizens,” while Republicans dispute that the voter registration and ID bill would block any legitimate voters. Election experts say the bill, which isn’t expected to pass, would make it difficult for some unknown number of voters to register and cast a vote.
At times, Schumer has used more definitive language about the bill’s impact, saying that “more than 20 million legitimate people … will not be able to vote under this law” or that it “would disenfranchise tens of millions of people.”
Walter Olson, a senior fellow at the Cato Institute’s Robert A. Levy Center for Constitutional Studies, told us the legislation wouldn’t meet the dictionary definition of “disenfranchise,” which is to “deprive a person of the right to vote.” But it would, as described by Democratic Sen. Patty Murray, “‘make it harder and more expensive for [many people] to [register and] vote,'” Olson said in an email. “That extra hassle and expense would mean that some citizens eligible to register and vote will in practice not complete the needed process even though the bill does not take away their legal right to register or to vote.
“How many eligible people will fail to complete the process? Any estimate is guesswork at this stage in part because it depends on factors that the bill itself leaves unspecified,” he said.
Schumer’s 20 million figure comes from an estimate of the number of voting age Americans who don’t have easy access to citizenship documents that the bill would require to register to vote. According to a 2023 survey by New York University’s Brennan Center for Justice and other groups, more than 9% of Americans of voting age, or 21.3 million people, wouldn’t be able to “quickly find” documents such as a passport, birth certificate or naturalization papers if they “had to show it tomorrow.” More than 3.8 million of those people don’t have those documents, the survey found.
That doesn’t mean that at least some of those Americans couldn’t obtain or find proof of citizenship in order to register to vote under the legislation. But some could find the process too onerous to complete, experts say. Under the bill, citizenship documents also would need to be presented in person to an election official if registering to vote for the first time or reregistering after moving, changing one’s name or making other changes to voter registration.
Eliza Sweren-Becker, deputy director of the voting rights and elections program at the Brennan Center for Justice, told us that “it’s definitely safe to say that millions of Americans would be blocked from voting” by the bill’s registration requirements, among other provisions. She noted that tens of millions of Americans register or update their registrations in the two years before elections. More than 103 million did so in the two years before the 2024 election, according to survey reports by the U.S. Election Assistance Commission.
“As many as 21 million could be stopped from voting” under the SAVE America Act, she said, because they lack ready access to a passport, birth certificate or naturalization document required under the bill for voter registration.
Schumer has repeatedly used the 20 million estimate, adding that these voters could be purged from the voter rolls and not know about it until they showed up to vote, at times linking this to a requirement under the bill for states to use a Department of Homeland Security database to remove noncitizens. “Our objection is it’s a voter suppression bill. Twenty million, maybe more people, when they show up to vote … will be told, you’re off the rolls. That’s the problem with the bill,” Schumer said in a March 17 press conference.
On the Senate floor the same day, the Democratic leader said, “It could purge millions of American citizens from the voter rolls through a screening algorithm designed by Elon Musk’s DOGE squad. It could disenfranchise over 20 million American citizens.”
The DHS database is known to have wrongly flagged as noncitizens some Americans who are, in fact, citizens. But the extent of those flaws is unclear — as is how voters might be notified and purged from voter rolls under the legislation.
Republican Sen. John Cornyn objected to Schumer’s remarks. On the Senate floor on March 19, Cornyn said that Schumer’s “general argument that American citizens would be denied the opportunity to vote is patently false. Thirty-eight states, including states like Georgia and Rhode Island, currently represented by Democrats, require voter ID. Are those states suppressing the vote? Is the minority leader suggesting that 38 out of our 50 states are actively engaged in voter suppression? Well, that is preposterous on its face.”
“So the idea that the SAVE America Act will disenfranchise legitimate voters is a bald-faced—well, let me try to be generous. It is not true, and he knows it,” Cornyn said, adding that Schumer was telling “people who may not be informed about the details of this that we are trying to take away their right to vote.”
Cornyn is nearly correct on the number: 36 states have some form of voter ID laws. But the requirements in the bill before the Senate are “stricter” than most of those state laws, according to the National Conference of State Legislatures.
We’ll explain what the bill would require for registering and casting votes, and how this could affect voters. (For more, see our article “Q&A on the SAVE America Act.”)
The SAVE America Act passed the House in February, and the Senate began debate on the bill on March 17. Similar legislation in recent years has failed to pass the Senate. A proposed Senate amendment would impose more restrictions on voting by mail, eliminating universal mail-in voting and only allowing mail ballots in certain cases, such as illness or disability, travel, or military service. Here, we describe the bill as it passed the House.
Republicans say the bill is needed to prevent noncitizens from voting in federal elections, though election experts say, and state audits have shown, this is rare.
Current federal law requires those registering to vote to attest that they are citizens under penalty of perjury. The SAVE America Act would require documentary proof, presented in person to election officials, for those registering or reregistering to vote.
This would happen “any time you conduct what we call a registration transaction, which usually comes from a life event, a move or a change of name,” David Becker, founder and executive director of the nonpartisan Center for Election Innovation & Research, which works with election officials throughout the country, said in a March 18 media briefing.
For most people, this would likely mean showing a U.S. passport or a certified birth certificate along with a driver’s license or government-issued photo ID. As we’ve explained, the bill stipulates elements the birth certificate must have, such as a government seal.
Some voters could prove citizenship with other documents. A REAL ID driver’s license doesn’t typically show citizenship, but five states issue REAL IDs that do. Also acceptable under the bill: a military ID and service record that says the person was born in the U.S., or a government-issued photo ID that shows a U.S. birthplace. Those with government-issued photo IDs that don’t indicate citizenship would also need either the certified birth certificate or a hospital birth record, adoption decree, a consular birth report, a naturalization certificate, or an American Indian card with the classification “KIC,” which designates U.S. citizenship for Mexican-born members of the Kickapoo tribes of Texas and Oklahoma.
As we said, surveys show millions of Americans could have trouble producing the proper citizenship documents. In addition to the 2023 survey Schumer has cited, the Bipartisan Policy Center, in analyzing the 2024 Survey on the Performance of American Elections, found that 12% of registered voters, the equivalent of 28.4 million citizens of voting age, lacked either a valid passport or a birth certificate they could easily find along with a valid government-issued photo ID.
For those who do have the proper documents, the requirement to show them “in person” could dissuade others from registering to vote. The bill says that people registering by mail won’t be registered unless they present “documentary proof of United States citizenship in person to the office of the appropriate election official.”
Sweren-Becker said that this in-person requirement would be “especially hard” for “working parents, people with disabilities, elderly voters, voters who live in rural areas.”
The bill calls for states to make unspecified “reasonable accommodations” for people with disabilities.
Republican Sen. Mike Lee said on the Senate floor on March 19 that claims about the legislation disenfranchising voters were wrong. “Ideally” Americans have the proper documents, he said, but “even if you do not have a single shred of documentation as to your citizenship — you can’t find it, it burned down, whatever it is — all you have to do is swear an affidavit.”
“The state is in a very good position to track down the details of the affidavit and easily confirm or refute what the person says,” Lee said.
The bill does provide a process for those who don’t have the required documents. It says: “Subject to any relevant guidance adopted by the Election Assistance Commission, each State shall establish a process under which an applicant who cannot provide documentary proof of United States citizenship … may, if the applicant signs an attestation under penalty of perjury that the applicant is a citizen of the United States and eligible to vote in elections for Federal office, submit such other evidence to the appropriate State or local official demonstrating that the applicant is a citizen of the United States and such official shall make a determination as to whether the applicant has sufficiently established United States citizenship for purposes of registering to vote in elections for Federal office in the State.”
The election official making that determination also would need to sign an affidavit “swearing or affirming the applicant sufficiently established United States citizenship for purposes of registering to vote.”
There’s a similar process for people whose names differ from their documents, such as married women who changed their names. They can provide “additional documentation” on the name discrepancy or sign an affidavit.
Olson said there’s uncertainty about these alternative methods of citizenship verification. Will they “be relatively easy and generous, accepting common sorts of documents and an uncomplicated sworn statement that most eligible persons will feel comfortable signing?” he asked.
States’ procedures will be governed by guidance from the Election Assistance Commission, the bill says, an independent agency that has two commissioners appointed by Trump and two appointed by former President Barack Obama.
“In short, we aren’t going to find out what the bill does on many key questions until after we pass it into law and the EAC begins issuing guidance,” Olson said. “One of the reasons I am critical of the bill is that I don’t believe we should be asked to take it on faith that the EAC will issue practical guidance in good faith. If the EAC is going to issue guidance that causes an uproar because it sets requirements many legitimate voters cannot meet, we should know that now, not later.”
Sweren-Becker said that the affidavit method “is only available if a state or local election official deems that the registered has sufficiently established U.S. citizenship … so it leaves an enormous amount of discretion in local and state election officials’ hands.” The bill also would impose criminal penalties and civil liability on election officials who register someone “who fails to present documentary proof of United States citizenship,” the legislation says. “So in practice,” she said, election officials “will face a lot of pressure to construe it [the affidavit method] very, very, very narrowly out of rightful concern about their own liability,” Sweren-Becker said.
Becker, in the March 18 briefing, said the legislation “would incredibly negatively impact voters across the political spectrum. … I don’t think there’s anyone who can say definitively, if this were to pass, which party would be hurt more by it,” he said. “I think it’s highly likely that Republicans would likely be more hurt” than Democratic voters, “because a lot of the voters who have difficulty digging up their documentary proof of citizenship are Republicans.”
In pushing back on Schumer’s comments about disenfranchisement, Cornyn spoke about the bill’s photo ID requirements for casting a vote. “Thirty-eight states, including states like Georgia and Rhode Island, currently represented by Democrats, require voter ID,” he said.
As we said, 36 states do have some form of voter ID laws, but the SAVE America Act is “stricter” than most of them, according to the National Conference of State Legislatures.
The Republican bill would require “a valid physical photo identification” in order to cast a ballot in person. Those voting by mail would need to submit a copy of a photo ID, or the last four numbers of their Social Security number and an affidavit saying that they couldn’t obtain a copy of their ID.
A valid photo ID under the bill includes: a state-issued driver’s license or ID card issued by the motor vehicle agency that includes a photo and expiration date, a U.S. passport, a military ID, or a photo ID issued by a tribal government that includes an expiration date. There are exceptions for overseas uniformed services members and those who have the right to vote absentee via the Voting Accessibility for the Elderly and Handicapped Act.
The NCSL said most states’ laws are less strict. “Currently, each state determines the types of ID acceptable to vote, and that often includes student IDs, hunting and fishing licenses or other state-specific identification cards,” it said in a post on its website updated in March.
Thirteen states also accept non-photo identification, such as a bank statement. NCSL classified 10 of the voter ID states as having “strict photo ID” laws.
Georgia is one of them, but it still accepts a broader range of documents than the SAVE America Act would. Georgia accepts a student ID from a public college in the state, an expired state driver’s license, an employee photo ID from a government entity, or a free voter ID card issued by the state, among other documents, the Georgia Secretary of State’s office explains. To get an absentee ballot, a voter submits the number on a driver’s license or state-issued ID card, or a photo or copy of another listed ID, or a document that shows a name and address, such as a utility bill, bank statement or paycheck.
NCSL puts Rhode Island in its “non-strict photo ID” category, along with 13 other states. Rhode Island also issues free voter ID cards and accepts “ID issued by a U.S. educational institution,” the state Board of Elections says. No ID is required to cast a ballot by mail.
When we asked Cornyn’s office about his comments, a spokesperson pointed to some of his other remarks, including a March 19 post on X, which said: “These tactics are nothing more than fearmongering by Dems who are objecting to this because they want to make it easier for people to cheat. In a country with citizens bright enough to put a man on the moon and to build the strongest, most powerful military & the greatest economy the world has ever known, Americans are smart enough and capable enough to be able to locate their driver’s license when they cast a ballot and to establish their citizenship in order to qualify to vote. Any suggestion to the contrary is ridiculous.”
Schumer also objected to the bill’s provision requiring states to submit their voter rolls to DHS’ Systematic Alien Verification for Entitlements program and remove noncitizens from their rolls. The legislation “could purge millions of American citizens from the voter rolls,” Schumer said in the March 17 press conference. He later added: “Our objection is it’s a voter suppression bill. Twenty million, maybe more people, when they show up to vote … will be told, you’re off the rolls.”
On the Senate floor that same day, he repeated the idea that people could be removed from voter rolls and not know about it until they try to cast a vote. “The way this works, you don’t have to be notified if you’re kicked off the rolls. You show up on Election Day and they say, ‘We’re sorry Mr. Smith, Ms. Jones, you’re not on the rolls anymore.’ And then they make it impossible to re-register. Certainly, on that day you lose your right to vote,” the senator said.
In March 15 remarks, he said the bill’s requirements for states to use the DHS system “will purge tens of millions of people from the voter rolls. Once purged, you don’t even know it.”
There are a couple of provisions regarding purging voters. The first requires states to use the DHS system “for the purposes of identifying individuals who are not citizens of the United States and taking the necessary steps to remove such individuals who are not citizens from the official list, after notice is given to such individuals and such individuals are given the opportunity to provide documentary proof of United States citizenship.” As we’ve explained, the DHS system has been shown to have flaws and has wrongly identified people as being noncitizens.
When we asked Schumer’s office about the language in the bill, a spokesperson said the bill included “a requirement that they [voters] be told they have been flagged,” but no requirements about what form the notice would take or the “length of time” people would be given to respond. And there’s “no language in the bill about notice to the voter that they have been purged,” the spokesperson said.
The bill doesn’t provide more details on how states should give “notice” and an opportunity to dispute incorrect information before removing people from the rolls; nor does it say people should be notified again before being purged.
There’s another provision in the bill that says states could remove someone “at any time.” It says: “A State shall remove an individual who is not a citizen of the United States from the official list of eligible voters for elections for Federal office held in the State at any time upon receipt of documentation or verified information that a registrant is not a United States citizen.” That provision doesn’t say anything about a notice given before removing someone.
Election experts told us there’s ambiguity in the bill regarding these provisions. We reached out to the offices of Sen. Lee and Rep. Chip Roy, the authors of the legislation, about this issue, but we haven’t yet received a response.
“[I]t’s not obvious that all of the ways people will be removed from the rolls will be subject by the SAVE Act to notice and an opportunity to respond,” Justin Levitt, a professor of constitutional law at Loyola Marymount University’s law school, told us in an email. “I’d think there are constitutional protections that would kick in, but they’re not explicit in the statute, and that’d take someone litigating.” Levitt, who briefly was a White House senior policy adviser on voting rights during the Biden administration, said the bill “seems to contemplate at least some people being kicked off the rolls without being told,” though this could be a mistake in the drafting of the bill.
“As for how many, it’s a question I can’t answer,” he said, explaining that it depends on the accuracy of the SAVE database and how the process of comparing voter rolls works.
Olson told us that the provision on using the DHS SAVE system “appears to establish protections (notification and a chance to contest removal by supplying documents)” for voters flagged for removal under that system. But “some other persons removed from the voter rolls may not have rights to notification and challenge unless their states have separately legislated to provide such rights,” he said, pointing to the provision on states removing noncitizens “at any time.”
“So far as I can tell, this means that anyone, including the federal government or some private person or group, can send ‘documentation or verified information’ to a state that a certain person, or a list of persons, on its voter rolls are not U.S. citizens. The state then ‘shall’ remove them,” Olson said. “So long as this is not being done by the method carved out for the SAVE database and its intersection with state voter rolls in federal possession, I don’t see where the bill provides any assurance of notification.”
Sweren-Becker had the same reading of the bill. “Absolutely, I think that the second provision … indicates that people could be removed, but on the basis that something has flagged them as a noncitizen, without notice to the voter or an opportunity to provide evidence of their citizenship,” she told us. “And it is also important to note that it is very unclear what ‘documentation or verified information’ means” and from what sources. “I think there’s a risk that election officials may receive, essentially, purge lists generated by activist groups who are not doing careful list matching.”
As for how many legitimate voters could be removed from voter rolls through this process, “I don’t know how to hazard a guess there,” Sweren-Becker said, noting that “shoddy” purge lists by activist groups have listed thousands of people.
Schumer, however, has gone as far as saying that, under the bill, 20 million could be wrongly purged without knowing they were removed from the voter rolls. But that figure comes from the estimate of those lacking easy access to a passport, birth certificate or naturalization papers. It’s not an estimate of voters who could be purged without their knowledge.
On March 17, the Senate began debate on the SAVE America Act, a Republican-backed voter identification and registration bill that passed the House last month. Here, we answer several questions about the legislation, many of them asked by our readers.
Previous versions of the bill, called only the SAVE Act, died in the Senate, where the measure hasn’t garnered 60 votes to overcome a filibuster and force a final vote. The new legislation could well face a similar fate — eventually — but the Republican leadership is holding a weeklong (or so) debate in an effort to attract support.
David Becker, founder and executive director of the nonpartisan Center for Election Innovation & Research, which works with election officials throughout the country, said in a March 18 media briefing that it was “extremely unlikely, if not impossible, that this passes.” He predicted that “next week, we’re not going to be talking about this.”
But this week, the Senate is going to be talking about it a lot. On the opening day of debate, Senate Majority Leader John Thune called the bill “a package of commonsense measures” that was about “ensuring that those who are registered to vote are eligible to vote – and that those who show up to vote at polling places are … who they say they are.” Senate Minority Leader Chuck Schumer called it “in every sense a voter suppression bill” that could “disenfranchise” millions of American citizens.
The SAVE America Act (or Safeguard American Voter Eligibility Act), passed the House on Feb. 11. The bill aims to prevent voting in federal elections by people who aren’t U.S. citizens — something that election experts say is a rare occurrence. Unlike last year’s SAVE Act, the bill also would require voters to present photo identification before casting a vote, whether by mail or in person. And states would have to use a Department of Homeland Security system to check the citizenship status of people on their voter rolls.
President Donald Trump has demanded that other measures be added to the legislation, including abolishing most mail-in voting.
We’ll explain more about the bill below.
There’s no requirement in the bill for all registered voters to reregister. However, if a voter did need to reregister for other reasons, such as moving or changing their name, they would have to show documentation proving their citizenship. “Under any method of voter registration in a State, the State shall not accept and process an application to register to vote in an election for Federal office unless the applicant presents documentary proof of United States citizenship with the application,” the legislation says.
Ceridwen Cherry, legal director of VoteRiders, a nonpartisan group that helps people get an acceptable form of identification so they can vote, told us that “any change to the registration would require documents to prove citizenship under the SAVE America Act. The statute is drafted broadly enough to encompass all changes to registration.”
VoteRiders’ mission is “to eliminate ID barriers to the ballot box so every eligible voter can cast a ballot that counts,” and as such, it opposes this legislation.
Becker, who said the legislation would “expansively … alter voting in every single state,” costing “tens, perhaps hundreds of millions of dollars,” said voters would need to prove citizenship under the bill “any time you conduct what we call a registration transaction, which usually comes from a life event, a move or a change of name.” (He also said that “in talking with election officials across the country, I have yet to find really any election official who supports this on either side of the aisle. It would make their jobs extremely more difficult” while primaries are occurring and months away from the general midterm elections.)
Current federal law requires those registering to vote to attest that they are citizens under penalty of perjury. The SAVE America Act would require people to present citizenship documents in person to election officials, even if they are registering by mail.
For most Americans registering to vote, proving citizenship would mean presenting either only a U.S. passport, or a certified birth certificate along with a driver’s license or other government-issued photo ID. The legislation lists requirements the birth certificate must meet, such as including the full names of at least one parent, the signature of an authorized government official, and the seal of the state or local/tribal government that issued it.
The Bipartisan Policy Center noted in a March 16 post that not all birth certificates include all of the criteria. About 53% of the U.S. population has a U.S. passport, according to Department of State data.
These are other types of documents besides a passport that would suffice to prove citizenship under the bill: a REAL ID driver’s license that indicates citizenship (five states have such “enhanced” driver’s licenses that include citizenship); a military ID and service record that says the person was born in the U.S.; or a government-issued photo ID that shows a U.S. birthplace. If presenting a government-issued photo ID that doesn’t say the person was born in the U.S. or has citizenship, a registrant would also need either the certified birth certificate or a hospital birth record, adoption decree, a consular birth report, a naturalization certificate, or an American Indian card with the classification “KIC,” which designates U.S. citizenship for Mexican-born members of the Kickapoo tribes of Texas and Oklahoma.
The Bipartisan Policy Center analyzed the 2024 Survey on the Performance of American Elections conducted by the MIT Election Data + Science Lab and found that 12% of registered voters lacked either a passport or a birth certificate along with a government-issued photo ID — the most common ways people would prove citizenship under this bill. The analysis also found that “wealthier and more highly educated voters are more likely to have documentary proof than others.” It found that “registered Democrats are more likely to have a valid passport than registered Republicans” and “Republicans are more likely to have a birth certificate than Democrats.”
According to a 2023 survey by New York University’s Brennan Center for Justice and other groups, more than 9% of Americans of voting age, or 21.3 million people, didn’t have easy access to citizenship documents, meaning they wouldn’t be able to “quickly find” such documents if they “had to show it tomorrow.” The percentage was 11% for Americans who did not identify as white.
In a summary of the bill, the nonpartisan Congressional Research Service explains that if people lack valid documents, “the bill would require states to establish a process whereby applicants could submit other documentation and sign an attestation under penalty of perjury that the applicant is a U.S. citizen and eligible to vote in federal elections.” If the person lacks documentation, the bill also would require the election official to sign an affidavit saying the registrant sufficiently demonstrated citizenship.
We received several questions from readers who are married, or divorced, and have changed their names, asking about how they can prove citizenship and ensure they can vote, should this bill become law. We wrote about these concerns last year as well. The bill includes a provision on name discrepancies, requiring states to establish a process for those registrations. (Again, voters who are already registered wouldn’t need to prove citizenship under legislation unless they needed to reregister.)
Cherry, with VoteRiders, told us that “if a voter has experienced a name change they would not be able to use their birth certificate as their only proof of citizenship as this document does not get updated if someone changes their name through marriage or divorce. They also could not use any of the other listed documents (e.g. passport) as their sole proof of citizenship if their name on the document does not match their current legal name.”
The bill requires states to set up a process to accommodate this. “Voters will either be able to provide ‘additional documentation as necessary to establish that the name on the documentation is a previous name of the applicant’ or ‘an affidavit signed by the applicant attesting that the name on the documentation is a previous name of the applicant,'” Cherry said. “The bill text does not lay out exactly what this process will be or what additional documentation would be accepted. It also leaves open the possibility for inconsistent rules between states.”
In general, the bill calls for the federal Election Assistance Commission, an independent, bipartisan agency, to issue guidance to states on implementing the legislation within 10 days of its enactment.
When we wrote about the SAVE Act last year, Wendy Weiser, vice president for democracy at the Brennan Center, raised concerns about criminal penalties in the bill for election officials. That provision remains in this year’s legislation. Weiser told us, “Any state process would be severely undercut by another provision in the bill making it a federal crime for election officials to register anyone who does not present ‘documentary proof of citizenship.’ How many election officials would be willing to risk incarceration and steep fines to register someone whose documentation does not match their current name?”
In a statement to us last year, Republican Rep. Chip Roy of Texas, who introduced the SAVE Act in the House and this year’s SAVE America Act, said concern over married women not being able to register to vote was “absurd armchair speculation.” He said the bill “provides a myriad [of] ways for people to prove citizenship and explicitly directs States to establish a process for individuals to register to vote if there are discrepancies in their proof of citizenship documents due to something like a name change.”
New in this year’s legislation is a nationwide voter photo ID requirement. Those voting in person would need to present “a valid physical photo identification” in order to cast a ballot. Those voting by mail would need to provide a copy of the photo ID.
Those who don’t have an ID for in-person voting could cast a provisional ballot and then would have three days to present their ID to election officials — or sign an affidavit “attesting that the individual does not possess the identification required … because the individual has a religious objection to being photographed.”
For by-mail voters, they also could submit the last four numbers of their Social Security number and an affidavit “attesting that the individual is unable to obtain a copy of a valid photo identification after making reasonable efforts to obtain such a copy.”
A valid photo ID for this purpose includes: a state-issued driver’s license or ID card issued by the motor vehicle agency that includes a photo and expiration date, a U.S. passport, a military ID, or a photo ID issued by a tribal government that includes an expiration date.
The National Conference of State Legislatures, which tracks state legislation, has said that these voter ID requirements “are stricter than those that exist in most states.” In a Feb. 19 post, NCSL staff wrote, “While 36 states currently have voter ID requirements to vote, state approaches vary. Just 10 states fall into the strict photo ID category, as defined by NCSL.”
An acceptable ID for these 36 states “often includes student IDs, hunting and fishing licenses or other state-specific identification cards.” Thirteen states accept non-photo identification, such as a bank statement. That’s broader than what the SAVE America Act would accept.
There are exceptions to the by-mail ID requirements for overseas uniformed services members and those who have the right to vote absentee via the Voting Accessibility for the Elderly and Handicapped Act.
We’ve written about this issue a few times. Last April, we explained that detailed audits of voting records by some states had found instances of noncitizens casting votes to be relatively rare. In some cases, officials in those states found hundreds of noncitizens on voter registration rolls, a fraction of whom also voted.
Noncitizens convicted of voting in federal elections face fines, jail time and deportation.
“The evidence is that the number of noncitizens illegally voting in federal elections is extremely low, not high enough to have changed the party outcome of any federal election in recent years,” Walter Olson, a senior fellow at the Cato Institute told us. “Audits and investigations in states like Ohio, Nevada, and North Carolina have found the numbers to be tiny in relation to votes cast. … The consistent experience has been that very few persons in this category mistakenly or deliberately vote.”
For instance, the Ohio Secretary of State announced in May 2024 that it found 137 people on the state’s voter registration rolls who had twice confirmed their noncitizenship status to the state motor vehicles bureau. The announcement didn’t say whether any had tried to actually vote. A grand jury indicted six people who legally and permanently immigrated to the U.S. for voting illegally as noncitizens between 2008 and 2020. In Georgia, a 2022 review found that 1,634 people had attempted to register to vote between 1997 and 2022 and could not be verified as citizens. None had voted. In October 2024, the Associated Press reported that Georgia election officials said 20 out of the 8.2 million on the state’s voter registration rolls were not U.S. citizens, and that nine had voted in previous elections.
The Bipartisan Policy Center analyzed a database of fraud cases compiled by the conservative Heritage Foundation and found “only 77 instances of noncitizen voting between 1999 and 2023.”
Last April, we were writing about unsupported claims from Elon Musk and the Department of Government Efficiency to have found evidence of large-scale voting by noncitizens. DOGE said it provided data to federal prosecutors for investigation. But nearly a year later, nothing has been made public about that investigation.
More recently, a systematic review of claims about noncitizen registrants and voters in all 50 states by the Center for Election Innovation & Research, updated in February, found that “sweeping allegations about noncitizen registrations or voting appear to arise from misunderstandings, mischaracterizations, or outright fabrications about complex voter data. In every examined case, when claims about large numbers of noncitizens on voting rolls are subject to scrutiny and properly investigated, the number of alleged instances falls drastically.”
Numerous states recently have used a Department of Homeland Security program called the Systematic Alien Verification for Entitlements, or SAVE, to check the citizenship status of people on their voter rolls — something that the SAVE America Act would require. The bill says that states should use the system “for the purposes of identifying individuals who are not citizens of the United States and taking the necessary steps to remove such individuals who are not citizens from the official list, after notice is given to such individuals and such individuals are given the opportunity to provide documentary proof of United States citizenship.” The legislation doesn’t provide more information on how these notices and opportunities to fix a mistake would be carried out.
Recent reporting shows the SAVE database has flaws.
According to a January New York Times article, 49.5 million voter registrations have been checked in several states, and the Department of Homeland Security referred about .02%, or 10,000 cases, to investigators. The Times found that when some counties began looking into the cases, it turned out that only a fraction of them were potentially noncitizens. There was no indication of how many of those who may have improperly registered to vote actually voted.
Texas, too, found there were errors in DHS’ SAVE database. In October, the state said the database identified 2,724 potential noncitizens in its voter rolls of more than 18 million people, and it referred the cases to Texas counties. Many of those counties found U.S. citizens were among those flagged.
In February, ProPublica and the Texas Tribune wrote that their examination of the SAVE system “reveals that DHS rushed the revamped tool into use while it was still adding data and before it could discern voters’ most up-to-date citizenship information.
“As a result, SAVE has made persistent mistakes, particularly in assessing the status of people born outside the U.S., data gathered from local election administrators, interviews and emails obtained via public records requests show. Some of those people subsequently become U.S. citizens, a step that the system doesn’t always pick up,” the news organizations wrote.
Yes, according to a February Harvard CAPS/Harris poll, which found that 71% of the registered voters surveyed said that they supported the SAVE America Act, including 91% of Republicans, 69% of independents and 50% of Democrats.
The online poll conducted Feb. 25-26 asked 1,999 registered voters, “Do you support or oppose the proposed SAVE America Act that would: Require proof of citizenship to register to vote, Require voter ID, Require states to remove non-citizens from their voting rolls, Require states to share unredacted voting rolls with the Department of Homeland Security.”
Three out of the four proposals mentioned in that description of the bill appealed to an even larger group. A press release about the results of the Harvard CAPS/Harris poll said, “The majority of voters support specific requirements of the Act, including proof of citizenship (75%), voter ID (81%), states removing non-citizens from voter rolls (80%), and states sharing redacted voting rolls with the Department of Homeland Security (61%).”
Past polls have revealed similar levels of support for some of those policies.
A Pew Research Center poll from August found that 83% of those asked were in favor of a requirement for everyone to show government-issued photo identification before voting, including 95% of Republicans and 71% of Democrats.
In addition, a Gallup poll from October 2024 found that 84% of surveyed adults supported “[r]equiring all voters to provide photo identification at their voting place in order to vote,” while 83% backed “[r]equiring people who are registering to vote for the first time to provide proof of citizenship.” About two-thirds of Democrats supported both ideas, more than 8-in-10 independents did, and nearly all Republicans were on board with each.
Becker, of the Center for Election Innovation & Research, noted that the results of these surveys depend on what questions are asked. “If you just ask the regular question in polls, do you support voter ID, you do see vast majorities of Americans say yes, including majorities of Democrats. If you ask people, should eligible voters without voter IDs be disenfranchised, you get very different responses.”
The Harvard CAPS/Harris poll also asked, “Which of the following is more important?,” giving two choices. A little more than half, 54%, said, “That we do everything possible to stop voter fraud and illegal immigrants from voting,” and 46% said, “That eligible citizens aren’t denied the ability to vote.”
Trump has proposed that the final version of the bill also eliminate mail-in voting with limited exceptions.
“We don’t want mail-in ballots,” Trump said while talking about his proposal during an interview with a Cincinnati news station on March 11. “We don’t want to have ballots coming from all different corners of the world. We want to have it accurate, and you can’t do that with mail-in ballots.”
In multiple posts on social media in March, the president has written, “NO MAIL-IN BALLOTS (EXCEPT FOR ILLNESS, DISABILITY, MILITARY, OR TRAVEL!).”
As is, the House-passed bill would not abolish mail-in voting, but it would require identification to both request and submit a mail-in ballot.
As we’ve reported, mail-in voting is used widely throughout the U.S. Eight states and Washington, D.C., conduct their elections mostly by mail, according to the National Conference of State Legislatures. In addition, 28 states allow “no excuse” mail-in voting, which means that voters don’t need to provide a reason when requesting a mail-in ballot.
In the August Pew Research Center poll, 58% of respondents said they supported allowing any voter to vote by mail.
Elections experts have told us for years that while fraud is slightly more prevalent with mail-in voting than in-person voting, it is still relatively rare and not widespread.
More recently, Trump has said that he wants the legislation to address two non-election-related issues.
“I added on no men playing in women’s sports, and I added in no transgender surgery, the mutilation of our children,” Trump said from the Oval Office on March 16, referring to his proposed ban on transgender women playing in women’s athletics and gender-affirming surgery for minors.
Those are the last two of Trump’s five-point plan for the bill, and Republican Sen. Eric Schmitt of Missouri has introduced an amendment to include all five parts in the final legislation.
“I’ve worked closely with President Trump and the White House to introduce a substitute amendment that will save our elections, save women’s sports, and save our children from gender mutilation surgeries. It’s time to get this done,” Schmitt said in a March 17 statement.
In all, Schmitt said his amendment would: “Require all voters to show ID,” “Require proof of citizenship to vote,” “End mail-in balloting with exceptions for military, illness, travel, and disability,” “Keep men out of women’s sports,” and “Protect children from transgender mutilation surgeries.”
Robert Farley contributed to this article.
Clarification, March 20: We made clear that the REAL ID “enhanced” driver’s licenses available in five states indicate citizenship.
Trump administration officials have claimed that they are saving Americans over $1.3 trillion by ending regulation of greenhouse gas emissions from cars and trucks. But the figure does not incorporate any benefits of the emissions standards. By one of the Environmental Protection Agency’s own calculations, getting rid of the standards could cost billions.
On Feb. 12, the EPA announced that it was revoking the 2009 endangerment finding, which allowed the agency to regulate greenhouse gases such as carbon dioxide that trap heat and cause climate change. Without that policy in place, the agency said it was terminating its rules that limit such pollution from vehicles. The regulation has primarily acted to increase fuel efficiency, since more efficient cars and trucks burn less gas and release less carbon pollution.
Since unveiling the finalized rulemaking last month that eliminates the emissions standards, officials have frequently touted an alleged savings of $1.3 trillion.
“This action will save American taxpayers over $1.3 trillion,” EPA Administrator Lee Zeldin said in the press conference announcing the policy change. “What that means is lower prices, more choices, and an end of heavy-handed climate policies. With today’s announcement, American families will save over … $2,400 for a new vehicle.”
In the same briefing, President Donald Trump and Russell Vought, director of the Office of Management and Budget, also mentioned the $1.3 trillion figure. White House Press Secretary Karoline Leavitt has similarly referred to it, and in official communications, the EPA has also emphasized it.
But the $1.3 trillion is not a net total. It only includes the added cost of making cars and trucks more fuel efficient over a period of nearly three decades, without considering any of the benefits, such as reduced fuel or maintenance costs. One of the agency’s own estimates, which also ignores any health or environmental benefits, shows that repealing the policy could ultimately cost Americans $180 billion.
“This is a very biased and misleading way to talk about the effects of this rollback,” Jason Schwartz, regulatory policy director at New York University School of Law’s Institute for Policy Integrity, told us of the $1.3 trillion framing. “It’s actually worse than only looking at one half of the equation because they’ve also left a bunch of really important effects off the other side of the equation.”
“I honestly can’t recall another rulemaking where the focus of all of the fact sheets and press releases was ONLY about the costs of the policy,” Kenneth Gillingham, a Yale University economics professor, told us in an email.
With the elimination of these standards, soon there will be effectively no fuel efficiency standards in place on American cars and trucks. While the Department of Transportation’s fuel economy standards still technically exist, with the passage of the One Big Beautiful Bill Act earlier this year, Congress set the compliance penalties in the program to zero. The Trump administration is also now working to finalize significant rollbacks to the standards.
Here, we’ll take a closer look at the EPA’s calculations and explain why the administration’s singular focus on $1.3 trillion is misleading.
According to an EPA fact sheet, the $1.3 trillion in alleged savings for U.S. consumers includes about $1.1 trillion in avoided vehicle technology costs and $200 billion in avoided electric vehicle charger and equipment costs between 2027 through 2055.
Those figures are taken from the agency’s regulatory impact analysis. The analysis modeled four different scenarios, making different assumptions about the future cost of gasoline and to what degree fuel savings are counted, under two discount rates. (A discount rate is how much future money is discounted on an annual basis to convert it to a present value.) For each of the scenarios at the 3% discount rate, the EPA estimated a “savings” of either $1.29 trillion or $1.34 trillion.
But those estimates are only one half of the EPA’s ledger. Partially or fully offsetting the alleged savings are a variety of increased costs as a result of undoing the standards, primarily due to higher fuel and maintenance costs (electric vehicles are cheaper to maintain). In no case is there a net savings of $1.3 trillion.
Most of the EPA’s modeled scenarios do show a net savings in repealing the greenhouse gas emissions standards. However, as we said, that analysis does not factor in any health or environmental benefits.
“A correct cost-benefit analysis – even under the Trump Administration’s Circular A-4 guidance document – must include all relevant benefits and costs,” Gillingham told us, referring to the Office of Management and Budget’s guidance for such analyses. “In this rule, they are simply arguing that the benefits don’t exist. The science is not on their side on this one.”
Even still, one of the agency’s key estimates — one using the Energy Information Administration’s best guess for future fuel prices — shows that rescinding the policy will cost Americans $180 billion on net. Even with nearly $1.3 trillion in savings, there will be almost $1.5 trillion in costs under that scenario.
The other scenarios assume either lower fuel prices or only factor in fuel savings for 2.5 years, or both — each of which lowers the benefits of any fuel efficiency policy and makes the standards appear more expensive.
In response to a series of questions about the agency’s calculations and why the agency is using the $1.3 trillion number when it is not a net figure, the EPA press office told us that the analysis included eight projected net impacts under different modeling assumptions (four scenarios under two discount rates). “We didn’t single out scenarios to suit a narrative, we followed the data,” the agency said in an email, which again highlighted the $1.3 trillion.
Although the agency and officials routinely refer to “over” $1.3 trillion in savings by ending the emissions standards program, only two of the EPA’s four scenarios at the 3% discount rate actually top $1.3 trillion in terms of the avoided higher technology costs that make vehicles more fuel-efficient. And at the 7% rate, those avoided costs never go above $870 billion.
“Rescinding the 2009 Endangerment Finding means real dollars back in the pockets of American families and unleashing consumer choice. Now, Americans will be able to buy the car they want, including newer, more affordable cars with the most up to date safety standards and that emit fewer criteria and hazardous air pollutants,” the agency continued, adding that the action “does not affect regulations on any non-GHG air pollutants.”
It’s true that the EPA retains its other regulations on vehicles that limit the release of so-called criteria air pollutants, such as ozone and fine particle pollution. But it’s not the case that ending the emissions standards will have no effect on those pollutants. Schwartz said the standards “would have had important indirect impacts on criteria pollutants.”
He cited an analysis by the Environmental Defense Fund, which estimated that with a repeal of the greenhouse gas emissions standards, there would be up to 58,000 more premature deaths and as many as 37 million more asthma attacks through 2055.
In its final rule, the EPA argued that removing the standards would have “only marginal and incidental impacts” on non-greenhouse gas emissions. The agency also said it “is no longer monetizing benefits” from reducing particle and ozone pollution because of uncertainty in how to do those calculations precisely.
“In the past, EPA has always considered air pollution benefits as part of a cost vs benefit sort of study. Here the assumption is that the pollution benefits are all zero,” Mark Jacobsen, an economist at the University of California, San Diego, told us, adding that the benefits are not just to the climate, but to air quality, which the EPA “has typically found to be quite important.”
In its 2024 rule finalizing increased greenhouse gas emissions standards under then President Joe Biden, the EPA estimated that those standards would provide $200 billion in health benefits due to less particle pollution and $1.6 trillion in climate benefits (in 2022 dollars, 3% discount rate). The overall net benefit was estimated at $2 trillion.
Even without factoring in health or environmental benefits, the EPA two years ago found a $62 billion net benefit to the increased standards, with higher vehicle technology and battery port costs more than offset by lower fuel, repair and maintenance costs.
This gets at problems several experts identified in the EPA’s analysis. Jacobsen told us that the agency’s estimates are “deeply flawed.”
Jacobsen, along with Gillingham and other economist and mechanical engineer co-authors, several of whom the EPA has frequently cited in its rulemaking, wrote a comment to the EPA explaining their criticisms of the agency’s proposed rule in September. While not all of the issues still apply to the revised final rule, some remain.
One of the biggest issues affects half of EPA’s scenarios, which only factor in 2.5 years of the standards’ fuel savings. In these scenarios, the fuel savings are dramatically reduced.
Gillingham, who said there were “many issues” with the EPA’s analysis, said this is “simply incorrect and flips it from how it has always been done.”
A major question for the cost-benefit analysis is why consumers undervalue fuel efficiency. As noted in the EPA’s regulatory impact analysis, some evidence suggests that buyers are only willing to pay for about 2.5 years’ worth of fuel savings upfront for a more fuel-efficient car. If it’s a market failure and buyers are simply not properly considering the benefits of better gas mileage for the lifetime of the car, then a cost-benefit analysis should factor in the fuel savings beyond the first 2.5 years. But if consumers dislike the features of fuel-efficient cars, then they could be already valuing and “paying” for those future fuel savings in the form of a less desirable vehicle, so those fuel savings should not be counted.
For half of its scenarios, the EPA assumes that the latter is true in full, so it doesn’t include fuel savings beyond the 2.5 years in its cost-benefit calculation.
“This makes a gigantic difference to the costs and benefits,” Gillingham told us.
Jacobsen, Gillingham and colleagues note in their comment that this assumption “is not supported by the economics literature.” Moreover, they said, the EPA is ignoring that some electric vehicle attributes are superior to conventional gasoline vehicles.
In its analysis, the EPA defended its approach on undervaluation, noting that it uses scenarios that both likely overestimate and underestimate the costs of the rule, and that it views them as “a form of a bounding exercise.”
Schwartz, of NYU, independently flagged the same concern about zeroing out most fuel savings as well. “They’re saying there must be $900 billion in lost vehicle features,” he said of the EPA. “That is absolutely not the case.”
At the same time, when the EPA estimates the cost of making vehicles more fuel efficient, the agency assumes that manufacturers preserve vehicle performance, and don’t opt for cheaper solutions, such as lowering a car’s horsepower, Schwartz said.
“By modeling both extra costs to maintain performance and assuming additional costs of alleged performance losses,” an economic report from Schwartz’s group explains, the EPA is “effectively double-counting costs.”
Gillingham and Jacobsen noted the same problem, calling it an “inconsistent combination” that makes the cost of regulation appear high and the benefits appear small.
Half of the scenarios also assume a “low” oil price to “take into account the policies being implemented by President Trump that are intended to drive down the price of gasoline and diesel,” according to the regulatory impact analysis.
But as Schwartz noted, and as we have explained before, oil is a global commodity. The price is set based on worldwide supply and demand. Economists and energy experts told us during the 2024 presidential campaign that Trump’s plans to reduce prices by increasing production are unlikely to be very successful over the long term. U.S. and global producers wouldn’t be incentivized to produce more if prices are low. In a world with less-fuel-efficient vehicles, demand will also be higher, driving up the cost of gasoline — a feature the EPA also did not fully account for in some scenarios, Schwartz said.
The various oil prices used are projections from the Energy Information Administration. Although the EIA also projects a “high” oil price, EPA did not use it for any of its scenarios.
“Every step of this analysis is biased in an effort to make their repeal look as favorable as possible and to hide from the public the real costs to consumers, to public health, and to the environment, that [are] going to result from this rollback,” Schwartz said.
Along with the $1.3 trillion, officials have often referred to an average per-vehicle savings of “over” $2,400 by eliminating greenhouse gas emissions standards. Zeldin gave this figure during the policy repeal announcement, and an agency fact sheet includes it.
In the announcement press conference, Trump also said that the repeal would “help bring car prices tumbling down dramatically. … You’re going to get a car that starts easier, a car that works better for a lot less money.”
But this figure, too, is not a net number and only captures the costs of the regulation, ignoring all benefits. The EPA’s estimate of $2,400 appears to comes from dividing the $1.1 trillion in reduced new vehicle costs by its estimated figure of how many new vehicles will be purchased.
As Gillingham told us, the savings “comes from fuel-saving technology not being installed. It definitely would not be reducing the price of a car by $2,400. It would mean that cars would not increase in price as quickly and that car buyers would lose out on a lot of fuel savings.”
The Food and Drug Administration on March 10 changed the approval for a version of the prescription drug leucovorin to include people with a very rare genetic condition. FDA Commissioner Dr. Marty Makary had previously implied that the drug’s new label would cover a much broader group of people with autism, saying that “hundreds of thousands of kids” would benefit.
The condition targeted in the FDA approval is a genetic version of cerebral folate deficiency, caused by mutations in a folate receptor gene. People with CFD — whether from genetic or other causes — have low levels of folate in their cerebrospinal fluid, which leads to reduced folate in the brain. This affects brain development. Patients with genetic CFD can experience developmental delays, movement disorders and seizures. Some behaviors are similar to those with autism.
However, this form of genetic CFD is estimated to occur in 1 in a million people, according to the FDA. That would translate to around 70 kids in the U.S. — far from “hundreds of thousands of kids.” Leucovorin had already been used for decades to treat genetic CFD via off-label prescribing, a common practice when evidence shows a drug approved for one condition improves another.
Despite this limited approval, Makary had initially implied a more substantial change. “Today the FDA is filing a Federal Register notice to change the label on an exciting treatment called prescription leucovorin so that it can be available to children with autism,” Makary said in a Sept. 22 press conference. “We are going to change the label to make it available,” he went on to say. “Hundreds of thousands of kids, in my opinion, will benefit.”
This was the same press conference in which President Donald Trump and others touted an unproven link between autism and the use of Tylenol, or acetaminophen, during pregnancy.
Makary later referred to a subset of people with autism with antibodies that block their own folate receptors, called autoantibodies. Some researchers have hypothesized that a subset of people with autism have CFD caused by these autoantibodies, but this is not well-established, as we will explain.
The FDA “is approving prescription leucovorin for treatment of autistic children,” Dr. Mehmet Oz, administrator for the Centers for Medicare & Medicaid Services, said at the same event. Health and Human Services Secretary Robert F. Kennedy Jr. said the treatment “may benefit large numbers of children who suffer from autism.” He had previously vowed by September to identify “what has caused the autism epidemic.”
The Federal Register notice Makary referred to described data on the rare genetic form of CFD, however. The notice also stated that data on leucovorin for people who have symptoms with “autistic features” along with antibodies targeting the receptor “is limited” and that “additional studies are needed.”
The then-head of the FDA’s drugs division, Dr. George Tidmarsh, also subsequently clarified that the new indication was the rare genetic one. “We’re not proposing to approve leucovorin for [people with] the diagnosis of autism,” he told the autism publication the Transmitter in an interview for a story published Oct. 2.
When asked this week about the discrepancy between Makary’s earlier comments about broad benefits for kids with autism and the ultimate FDA approval for a rare genetic condition, a spokesperson from the Department of Health and Human Services told us that Makary previously had been talking about an antibody-related form of CFD, and not the rare genetic disorder.
“Dr. Makary was referring to cerebral folate deficiency — which can be caused by antibodies blocking folate receptors — rather than cerebral folate transport deficiency, which is caused by a specific genetic mutation,” the HHS spokesperson wrote in an email.
However, as we’ve said, the idea that a large subset of people with autism have CFD and can benefit from leucovorin has not been well-established.
“There is no substantive evidence that cerebral folate deficiency (CFD) plays a role in the pathogenesis of autism,” two researchers with expertise in folate and cancer treatment wrote in a January perspective in the New England Journal of Medicine. They also said that despite claims that antibodies against folate receptors play a role in autism, most experts consider this conclusion to be “inconclusive.” They added that the presence of the antibodies doesn’t necessarily mean that folate is low in the cerebrospinal fluid, which is the defining feature of CFD.
The new approval was for GSK’s Wellcovorin, a brand-name version of leucovorin that has long been off patent and that is no longer made by the company. Leucovorin remains available in generic versions. It is mainly used for cancer patients alongside certain chemotherapy regimens to reduce toxicity or to improve effectiveness.
While clarifying that Makary’s remarks about broad benefits applied to a different form of CFD, the HHS spokesperson also said that the rare genetic form of CFD “was the focus of the September announcement about this drug.”
But during the Sept. 22 press conference and subsequent media appearances, Makary repeatedly emphasized potentially sweeping benefits of the new leucovorin label.
“For many kids with autism, it will provide some improvement in their symptoms, and for some subset, marked improvement,” Makary said in a Sept. 22 NewsNation interview, urging people to talk to their doctors. “There are 2.5 million kids suffering, and I hope hundreds of thousands of them will see some improvement with this new treatment that we’re going to approve in about two to three weeks,” he went on to say.
“I think the biggest story today was that the FDA is taking action to make leucovorin available to kids with cerebral folate deficiency,” he told ABC News that same day. “That may be 20% to 50% of kids with severe autism, and they have a clinical improvement in studies.” In a Sept. 25 interview on C-Span, he gave an even larger estimate, saying “we are going to approve a drug called leucovorin for the treatment of autism” and that it “may help 50% or 60% of kids with autism.”
There is very limited evidence to support the assertion that wide groups of kids with autism could benefit, as we wrote in September. David S. Mandell, a psychiatry professor at the University of Pennsylvania Perelman School of Medicine and director of the Penn Center for Mental Health, told us then that the evidence on leucovorin “as a treatment for autism is very weak.”
Other researchers told the Transmitter in September that the literature on autism and leucovorin was “meager” and that it would be “extremely premature” for the administration to recommend the treatment for autism.
“These leucovorin studies are small, lack validated biomarkers or outcome measures, and certainly are not generalizable to all children with autism,” Dr. Shafali Jeste, a neurologist at UCLA, told the Transmitter. “The over-simplified conclusions and media hype from these studies take advantage of vulnerable families who are searching for answers and hope.”
At the time, this evidence included a small collection of studies that looked at the impact of leucovorin on communication and other characteristics in children with autism. One of these studies — among the largest, with 80 participants recruited — has since been retracted due to concerns about its data and statistical analysis, according to a notice on the journal website. Another of the studies had been terminated for “investigator non-compliance,” although the authors still published results.
“Larger, well-designed, multisite trials using objective outcome measures are necessary to determine whether leucovorin is safe and effective in autism and in which subgroups it may be most beneficial for,” says an FAQ page from the American Academy of Pediatrics.
Despite these uncertainties and the lack of a broad approval, people appear to have heeded Makary’s advice to talk to their doctors about leucovorin. New outpatient prescriptions of the drug increased by 71% in children ages 5 and older in the first couple of months following the September announcement, according to a study published March 5 in the Lancet.
President Donald Trump claimed that Iran essentially shutting down the Strait of Hormuz “doesn’t really affect” the United States the way it does “other countries.” It’s true that a small share of U.S. oil imports comes from the Persian Gulf. But the U.S. has been affected by the global increase in the price of oil.
Since the waterway has been effectively closed – significantly reducing crude oil exports from the Persian Gulf region – oil prices have increased by double-digit percentages, which has contributed to a 50-cent-plus spike in the average price of a gallon of gasoline in the U.S.
“The US is definitely affected,” Mark Finley, the nonresident fellow in energy and global oil at Rice University’s Baker Institute, told us in an email. Because it’s a global oil market, “if something goes wrong anywhere, the price goes up everywhere,” he said.
Iran has blocked the flow of oil and other goods through the strait in retaliation for joint U.S. and Israeli airstrikes that began on Feb. 28. Iran has threatened to shoot or bomb vessels that attempt to pass through the narrow body of water that separates Iran from Oman and connects the Persian Gulf with the Gulf of Oman and the Arabian Sea.
About 20 million barrels per day of crude and other oil products were transported through the strait in 2025. That has slowed “to a trickle” since the U.S.-Israeli conflict with Iran began, according to the International Energy Agency.
In a March 9 press conference, Trump talked about offering “risk insurance” to oil tankers operating in the region, possibly by having U.S. Navy ships escort the tankers, “because you have to keep the straits flowing.”
But then he said, “With all of that, it affects other countries much more than it does the United States. It doesn’t really affect us. We have so much oil. We have tremendous oil and gas, much more than we need.” And he added, “I mean, we’re doing this for the other parts of the world, including countries like China. They get a lot of their oil through the straits. So, we’re doing this.”
Compared with some other nations, the U.S. does get just a fraction of its crude oil from Middle Eastern countries for whom the strait is the primary route of exporting oil products.
Last year, the U.S. imported approximately 490,000 barrels of crude oil per day from countries in the Persian Gulf, which include Bahrain, Iran, Iraq, Kuwait, Qatar, Saudi Arabia and the United Arab Emirates. That was about 8% of the almost 6.2 million barrels per day the U.S. imported in total, according to the U.S. Energy Information Administration. (Canada and Mexico were the source of roughly 70% of U.S. crude imports last year, with Canada alone accounting for a little more than 63%.)
Meanwhile, “About 80% of oil and oil products transiting the Strait in 2025 was destined for Asia,” the IEA has reported – with China, India and Japan being the main importers in the region. China, which Trump mentioned by name, receives between 45% and 50% of its imports through the strait, according to the Center on Global Energy Policy.
But it’s not accurate to claim that Iran’s blockade on the strait “doesn’t really affect” Americans, as Trump claimed. The fact that the U.S. is the world’s leading producer of crude oil, and relatively little of its imports come from the Persian Gulf, doesn’t mean that Americans won’t feel any pain.
“It insulates us in the sense that we’re not going to have a hard time finding supply, but the prices are global, so prices go up anyway,” Abhi Rajendran, director of Oil Markets Research at Energy Intelligence, told us in an interview.
As we’ve reported, the U.S. still relies on some imports because much of the crude oil produced domestically is lighter, or less dense, while many refineries in the U.S. were long ago configured to use the heavier crude oils produced in other parts of the world, such as Canada. That’s also why the U.S. exports a lot of the oil produced in the States.
Trump wrote on social media that higher oil prices are actually a positive thing. “The United States is the largest Oil Producer in the World, by far, so when oil prices go up, we make a lot of money,” the president said.
In his email, Finley told us that “US oil companies, their employees, and the states where they operate benefit from higher prices.” As for consumers, including households and businesses, he said they “bear the burden of higher prices at the pump” as well as on “everything that uses oil.”
He noted that the price of gasoline, diesel fuel and other petroleum products in the U.S. “have gone up sharply” since the U.S. and Israel attacked Iran.
In a February update, the IEA said, “With around 25% of the world’s seaborne oil trade transiting the Strait, and options to bypass it being limited, any disruption to flows through the Strait would have huge consequences for world oil markets.” It warned that a prolonged disruption of shipments would lead to oil supply shortages and make price increases inevitable.
As we’ve written, most of the cost of gasoline is determined by the price of crude oil, which refiners use to make gasoline and other petroleum products. The price of crude oil is set internationally and is largely based on supply and demand factors around the world.
Since the airstrikes on Iran began, the price of West Texas Intermediate crude, the U.S. benchmark, has increased about 41% to almost $95 a barrel, and the price of Brent crude, the international standard, rose about 32% to just over $94 a barrel, according to the Energy Information Administration. As a result, as of the week ending March 9, the average U.S. price for regular grade gasoline had increased to $3.50 per gallon – up by about 56 cents, or roughly 19%, since the week ending Feb. 23, which was five days before the fighting with Iran began, EIA data show.
On March 11, “to address disruptions in oil markets stemming from the war in the Middle East,” the 32 nations that are members of the International Energy Agency — which include the U.S. — announced that they collectively would make 400 million barrels from their oil reserves available for purchase “over a timeframe that is appropriate” for each country.
For the U.S., the Department of Energy said that Trump had authorized the release of 172 million barrels from the Strategic Petroleum Reserve over several weeks.
Experts have said that whether the releases help stabilize oil markets and lower prices depends on how fast the crude can be shipped and how much longer the fighting lasts.
“I think it will help,” Rajendran told us about the planned releases. But he added this caveat: “as long as the conflict doesn’t drag on past early to mid-April.”
Beyond that point, he said, countries would likely have to keep drawing more from their oil reserves or start making other adjustments to address demand.
Is the U.S. at “war” with Iran? Americans are getting conflicting messages from the Trump administration and congressional leaders.
“We are not at war. We have no intention of being at war,” House Speaker Mike Johnson said at a press conference on March 5, hours after Republicans in the House blocked a war powers resolution that would have required congressional approval for any further military action against Iran. Instead, Johnson called the military action a “limited operation.”
But in remarks to reporters on March 7 — and on other occasions — “war” is exactly how President Donald Trump has described it.
“We’re winning the war by a lot,” Trump told reporters on March 7. “The war itself is going unbelievably. It’s as good as it can be.”
While there are varying definitions of war even among academics who study such things, the war-or-not political debate is mostly about the legal definition of war according to the Constitution, and the implications that come with such a designation.
While Article II of the U.S. Constitution designates the president as “Commander in Chief of the Army and Navy,” Article I, Section 8 of the Constitution gives Congress — and only Congress — the power “To declare War.” In other words, the president is obligated to seek authorization from Congress before he initiates a war.
But Congress hasn’t formally declared a war since World War II. And it didn’t happen with the military attack initiated by Trump in Iran. Rather, in accordance with the 1973 War Powers Resolution, Trump provided a report to Congress on March 2 about the administration’s justification for the U.S.-Israeli joint strikes against Iran initiated on Feb. 28.
“So currently, if political leaders were to say that this is a war, they would also be acknowledging that the administration’s actions were unconstitutional,” Stephanie Savell, director of Brown University’s Costs of War project, told us.
In a March 1 post for his Substack, Foreign Exchanges, journalist Derek Davison wrote that Trump had “made a little verbal slip” when referring to the military operation as a war.
“You’re not supposed to refer to these sorts of things as ‘wars’ when you’re the president of the United States, at least not at their outset, because by law wars have to be declared by Congress,” Davison wrote. “Presidents have leeway to engage in military action prior to a congressional vote but only in self-defense, which was plainly not the case here even if one were to stretch that term beyond all comprehension.”
But Trump numerous times has referred to the situation with Iran as a war.
“We have unlimited middle and upper ammunition, which is really what we’re using in this war,” Trump said in remarks on March 3.
“We’re doing very well on the war front, to put it mildly, I would say,” Trump said on March 4.
In his remarks on March 7, when talking about American casualties, Trump commented, “It’s part of war. It’s a sad part of war. It’s the bad part of war.”
Secretary of Defense Pete Hegseth has also repeatedly referred to the armed conflict as war.
“We didn’t start this war, but under President Trump, we are finishing it,” Hegseth said in a press conference on March 2. “We set the terms of this war from start to finish.”
Those characterizations are in stark contrast to the way many Republican members of Congress have described the military conflagration.
“Nobody should classify this as war. It is combat operations,” Republican Rep. Brian Mast said on CNN the day the U.S. and Israel initiated airstrikes on Iran.
In a press conference on March 3, House Minority Leader Hakeem Jeffries pointed to Trump’s own words to argue that the president “has unconstitutionally and illegally chosen to launch a war.”
“He’s describing it as a war,” Jeffries said. Hegseth “is describing it as a war. Other members of the administration are describing it as a war. And it’s a requirement under the Constitution that it’s members of Congress who make the decision as to whether to get us entangled in this kind of armed conflict.”
As we’ve written before, legal experts have told us that under an originalist interpretation of the Constitution, congressional approval for the use of military force against another country is required. But in practice, several presidents have launched military actions in other countries without congressional authorization.
Robert Johnson, director of Oxford University’s Changing Character of War Center, told us via email, “There is a political reason not to call the campaign against Iran a war. The President must consult Congress and gain approval after 60 days. Until that time, he is permitted to take actions which are in self-defense of the United States, a power the POTUS was granted because [of] the Cold War and the speed at which a nuclear armed attack could be launched.”
“Most scholars and lawyers do not use the term war, even when they should,” Johnson said. “The term in use is armed conflict. This is further defined as an armed attack. A pattern has been set in the last three decades of not declaring war and taking military action, that is, using lethal force to obtain political ends and to neutralise an emergent threat, such as a terrorist attack. Legally, the criteria are that it should be a threat which cannot be dealt with reasonably by any other means and it should be ‘imminent’ as a threat.”
The media and academics, of course, use other definitions of war that have nothing to do with the legal or constitutional considerations.
The Associated Press, for example, decided on March 1 to start using the word “war” to refer to the Israeli-U.S. strikes on Iran and Iran’s retaliation.
“This reflects the scope and intensity of the fighting,” the AP wrote.
The AP noted that the Merriam-Webster dictionary defines war broadly as, “A state of usually open and declared armed hostile conflict between states or nations,” or “a state of hostility, conflict, or antagonism.”
“Even though none of the countries have officially declared war, the attacks by the United States and Israel, combined with Iran’s retaliation, meet those criteria,” the AP noted. “The decision by the Trump administration and Israeli leaders to attack and the subsequent destruction and casualties are enough to call the actions, and Iran’s response, a war. Trump himself has used the word war to describe the conflict.”
Johnson, of the Changing Character of War Center, said, “As a phenomenon, war is a contest of organised polities using lethal armed force at scale. Under this definition, the U.S. is ‘at war.'”
Savell, at the Costs of War project, cited the words of Douglas Fry, an anthropologist of war, in his 2007 book “Beyond War: The Human Potential for Peace.” Fry defined war as: “A group activity, carried on by members of one community against members of another community, in which it is the primary purpose to inflict serious injury or death on multiple nonspecified members of that other community, or in which the primary purpose makes it highly likely that serious injury or death will be inflicted on multiple nonspecified members of that community in the accomplishment of that primary purpose.”
“This fits what the US is doing in Iran,” Savell said.
But there are other definitions used in academia as well.
Scott Wolford, a professor of government at the University of Texas at Austin, and Jeff Carter, a professor in the Department of Government and Justice Studies at Appalachian State University, are co-directors of the Correlates of War Project, which provides a “systematic accumulation of scientific knowledge about war” dating back to 1816.
COW defines war as “‘sustained combat’ between belligerents, or what we might call competitive violence used by groups organized for violence against other groups organized for violence,” Wolford and Carter told us via email.
The conflict between the U.S. and Iran meets their definition of “sustained combat,” they said.
“Operationally, though, to enter the COW data as a war (as opposed to lower-level violence) there’s a battle death cutoff of 1000, above which a conflict enters the data as a war,” they said.
Seven American troops have been killed in the military conflict so far, and retaliatory Iranian strikes have also killed nearly two dozen others in the Middle East region, according to a March 8 New York Times report. Iran’s U.N. ambassador said on March 6 that more than 1,300 Iranian civilians have been killed in the conflict.
Those figures from Iran have not been verified, however, and Carter noted that COW’s 1,000 threshold “applies to members of the combatants’ armed forces,” not civilians.
If the military conflict leads to 1,000 battle deaths, it would be categorized as a war in the COW database, regardless of what either Iranian or U.S. leaders call it. (Iran’s foreign minister, Abbas Araghchi, did call the conflict a “war,” telling PBS News on March 9, “This is a war imposed on us.”)
“The virtue of those definitions is that they’re independent of what governments *say* about whether or not they’re at war,” Wolford and Carter wrote.
“But that’s distinct from the political-legal question of whether this is a war,” they said. “Declarations of war are pretty rare, though Congressional authorizations for the use of force aren’t, and the fact that this conflict began and continues with neither is probably what’s at issue in the public argument over the definition.”
But experts told us the political classification of the conflict could change over time, if the number of American casualties rose, if ground troops were deployed, or if the military action continues for a protracted amount of time.
“If there was a specific and limited set of armed attacks, of short duration, the Administration could sustain the argument that they are not yet at war,” Johnson said. “However, the scale, extent, and possibly duration of [counter] attacks would take us beyond purely legal definitions.”
In remarks on March 11, Trump referred to the military action in Iran as “a little excursion.”
A reporter asked, “You just said, ‘It is a little excursion,’ and you said, ‘It is a war.’ So which one is it?”
“Well, it’s both,” Trump said. “It’s both.”
President Donald Trump has claimed that the 2015 Iran nuclear deal was “a road to a nuclear weapon” and the country “would be sitting with a massive nuclear weapon three years ago” if he hadn’t withdrawn the U.S. from the deal in 2018 during his first term. The multilateral deal aimed to restrict Iran’s uranium enrichment program, and experts told us that after the U.S. withdrawal, Iran accelerated it instead.
It’s not possible to predict what would have happened if the agreement, called the Joint Comprehensive Plan of Action and negotiated by former President Barack Obama’s administration, had remained in place. In addition to imposing restrictions on Iran’s enrichment of uranium, the deal required international inspections of the country’s nuclear facilities.
On March 3, when speaking about the U.S. airstrikes on Iran that began Feb. 28, Trump said that Obama “made maybe the worst deal I’ve ever seen, because he gave all power in the Middle East to Iran, he went the exact opposite way, and I terminated that. If I didn’t terminate that deal, they would be sitting with a massive nuclear weapon three years ago, which would have been used already on Israel at least, and other countries also. And we wouldn’t be talking about it right now.”
The president went on to say that Obama “was giving them the right to have the path to a nuclear weapon,” saying that deal “expired.”
The next day, Trump said: “If we didn’t terminate the worst deal, one of the worst deals ever made, the Obama nuclear deal … it was a road to a nuclear weapon. Bad things would have happened four years ago, because they would’ve had a weapon four years ago, if I didn’t terminate that deal.”
And during a speech on March 11, Trump said, “But that deal, the Iran nuclear deal gave them the right to have a nuclear weapon as of three years ago.”
But several experts we spoke to disputed Trump’s claim and told us that Iran advanced its nuclear program after Trump’s decision to pull out of the agreement in his first term.
“Iran was able to advance its nuclear programme to the point where it was before the 12 Day War last June not because of the JCPOA, but because President Trump withdrew the United States from the JCPOA,” Laura Rockwood, senior fellow at the Vienna Center for Disarmament and Non-Proliferation, told us in an email. Rockwood worked for the International Atomic Energy Agency for 28 years, retiring in 2013.
Similarly, Richard Nephew, an international and public affairs senior research scholar at Columbia University who worked as a special envoy for Iran and for the State Department under the Biden administration, told us, “Trump’s decision to withdraw from the JCPOA in 2018 had a significant accelerating effect on the program.”
“The JCPOA would absolutely not have allowed Iran to develop nuclear weapons,” Nephew said. “First of all, there were prohibitions; then there were transparency requirements; and, then, there were the risks of snapback and punishment” if Iran violated the terms.
Daryl G. Kimball, executive director of the Arms Control Association, a nonpartisan organization that provides analysis on arms control and national security issues, told us for an earlier story that the 2015 nuclear deal “established an array of limits on Iran’s uranium enrichment and uranium stockpiling” and a rigorous monitoring and verification program. After the Trump administration’s withdrawal from the deal, “Iran began to reconstitute its nuclear capabilities, including by deploying large numbers of advanced centrifuges and stockpiling” highly enriched uranium.
As we’ve explained before, the nuclear agreement, which took effect in 2016 and was signed by the five permanent members of the United Nations Security Council — China, France, Russia, the United Kingdom and the United States — and Germany, restricted Iran’s ability to enrich uranium for 15 years and required monitoring and inspections of Iranian facilities for the same amount of time.
Under the deal, Iran agreed to do away with much of its nuclear program and, in exchange, the signatories lifted sanctions, the Council on Foreign Relations explained.
Trump announced on May 8, 2018, that the U.S. would withdraw from the deal and reinstitute sanctions. About a year later, in July 2019, Iran had exceeded the limits on its stockpile of low-enriched uranium that had been set in the JCPOA, the International Atomic Energy Agency reported at the time. Iran’s foreign minister said the country would begin to enrich uranium beyond the low level allowed under the deal (3.67%), which was the level needed for civilian nuclear power.
“The JCPOA dramatically restricted Iran’s ability to produce fissile material and, in particular, not only placed a cap on the quantity of enriched uranium Iran could stockpile and on the level of enrichment, but required the dismantlement of 2/3 of its centrifuges and limited its ability to produce advanced centrifuges,” Rockwood said. “Iran simply would not have been able to enrich to the point of possessing over 400 kg of 60% enriched uranium had the JCPOA remained in place.”
Rockwood was referring to the amount of 60% enriched uranium that Iran had stockpiled before the June 2025 U.S. bombing of nuclear program sites in the country. To be weapons-grade, the uranium would need to be enriched to 90%, as we’ve explained.
Of course, Iran could have violated the terms of the nuclear deal and pursued a nuclear weapon.
“No single element blocks Iran’s pathway to nuclear weapons, but taken together, the nuclear restrictions and monitoring form a comprehensive system that will put nuclear weapons out of Iran’s reach for at least 15 years,” the nonpartisan Arms Control Association explained in an August 2015 analysis. “Many of the JCPOA provisions also extend beyond 15 years. Monitoring of centrifuge production facilities continues for 20 years, and monitoring of uranium mines and mills continues for 25 years. International Atomic Energy Agency (IAEA) inspectors will have enhanced access indefinitely.”
Critics of the JCPOA — including Trump — have argued that the deal didn’t go far enough, and they objected to the lifting of economic sanctions.
“One of the main arguments used against the JCPOA was that it allowed Iran to continue enriching uranium and move closer to nuclear capability while remaining technically in compliance,” the nonpartisan Center for Arms Control and Non-Proliferation wrote in a June report. “The JCPOA also contained so-called ‘sunset provisions’ on various aspects of the deal such as lifting limits on centrifuges after 10 years or reduced enrichment beyond 3.67% only lasting for 15 years. This led to concerns that the deal would only temporarily delay Iran’s nuclear program while preventing parties from finding a more permanent solution. Additionally, critics worried that lifting sanctions on Iran in return for the JCPOA’s focus on constraining Iran’s nuclear program would diminish the United States’ ability to address other security concerns such as Iran’s missile program or its funding of violent non-state groups in the Middle East.”
In saying that Iran would’ve had a nuclear weapon “three years ago,” Trump may have been referencing one of these provisions, known as “transition day,” which was set to take effect on Oct. 18, 2023, eight years after implementation of the deal. On that day, if Iran had complied with its commitments under the deal, some of the restrictions on Iran’s nuclear and missile programs would have been lifted. However, while U.N. restrictions expired, countries that remained in the JCPOA after the U.S. withdrawal chose to maintain their restrictions, citing Iran’s noncompliance.
We asked the White House about Trump’s remarks, but we didn’t get a response.
While Trump claims that the JCPOA would have brought Iran closer to having a nuclear weapon and his withdrawal stopped that, the Center for Arms Control and Non-Proliferation estimated that the withdrawal sped up the so-called “breakout time,” or the time Iran would need to produce weapons-grade uranium that could then be used for one bomb – if the country chose to do so. The center estimated, as of November 2024, that the breakout time went from two to three months before the deal to 12-plus months during the deal. And then, after the U.S. withdrawal, the breakout time was reduced to just a couple of weeks.
As we’ve explained, it would take more time to actually develop a nuclear weapon. “After this point, once you have the weapons-grade uranium, Iran would then need to manufacture the rest of the weapon. This process would likely take much longer, perhaps months to a year,” Emma Sandifer, program coordinator at the Center for Arms Control and Non-Proliferation, told us for an earlier story.
She said that last June’s airstrikes likely lengthened the “breakout time,” but the IAEA hasn’t been able to inspect the damaged nuclear program sites since then.
In 2017, several months before withdrawing from the nuclear deal, Trump had claimed that Iran “has committed multiple violations of the agreement.” But as we wrote at the time, the IAEA said in its multiple reports after the deal went into effect that Iran was abiding by it. Trump himself had twice certified to Congress that Iran had complied with the deal, before claiming there had been violations.
In late September 2017, Gen. Joseph Dunford, then the chairman of the Joint Chiefs of Staff, told Congress that “Iran is adhering to its JCPOA obligations” and that the agreement “has delayed Iran’s development of nuclear weapons.”
Multiple news outlets have reported that video, satellite images and expert analysis indicate that the United States was likely responsible for the Feb. 28 bombing of an Iranian school for young girls, contradicting President Donald Trump’s unsupported claim that the deadly strike “was done by Iran.”
When a reporter aboard Air Force One asked Trump on March 7 if the U.S. had bombed the Shajareh Tayyebeh elementary school, the president said, “No, in my opinion, based on what I’ve seen, that was done by Iran.” He continued: “We think it was done by Iran – because they are very inaccurate, as you know, with their munitions. They have no accuracy whatsoever. It was done by Iran.”
Defense Secretary Pete Hegseth, who was standing near Trump at the time, didn’t echo the president’s version of events when a reporter asked if that claim was accurate.
“We’re certainly investigating,” Hegseth said, before adding that “the only side that targets civilians is Iran.”
But the available evidence suggests that Iran wasn’t at fault, according to several news reports.
The bombing happened on the first day of U.S. and Israeli airstrikes against Iran as part of the joint military mission known as Operation Epic Fury. The school was located in very close proximity to an Iranian naval base operated by the Islamic Revolutionary Guard Corps that was bombed in the air attacks. NBC News reported that the naval base had closed more than a decade ago, according to an official with Iran’s education ministry and a mother the network interviewed.
Iranian officials have said that more than 160 people, mostly students, were killed when the school was hit. But the number of casualties hasn’t been independently verified.
A video posted March 8 by the Mehr News Agency, which has been described as a semiofficial Iranian news service, shows a missile striking in the vicinity where the naval base and school were in southern Iran, according to news reports. Smoke was already visible in the surrounding area when the missile landed and exploded, creating a new, darker plume of smoke and debris. Multiple news organizations verified the video using geolocation tools.
The New York Times reported that satellite images it obtained from Planet Labs show “that multiple precision strikes hit at least six Revolutionary Guards buildings along with the school,” including four buildings that were completely destroyed. The Times, citing a timeline of the strikes, said that the video suggests that the school could have already been struck when that missile made impact with another structure.
The Washington Post reported that eight munitions experts said that the missile seen in the Mehr News Agency video, based on its shape, appears to be a Tomahawk Land Attack Missile, which the U.S. developed and is known to have used in its air assault on Iran. The U.S. military has released several videos and photos of those long-range missiles being launched from Navy warships during the now 11-day conflict.
Trevor Ball, a former U.S. Army explosive ordnance disposal technician who covers munitions for the investigative journalism group Bellingcat, wrote in a thread on X that the posted video “shows a US Tomahawk missile hitting an IRGC facility in Minab, Iran, on Feb 28, showing for the first time that the US struck the area.” He said, “The footage appears to contradict President Donald Trump’s claim it was an Iranian missile that hit the school.”
In a March 9 press conference in Miami, Trump still insisted that Iran could be responsible, saying it “also has some Tomahawks” and Iran “wish[es] they had more.” The president added: “But whether it’s Iran or somebody else, the fact that a Tomahawk, a Tomahawk is very generic.”
But there is no evidence that Iran has acquired Tomahawk missiles. “Iran has none, though it has lots of missiles of different kinds,” Mark Cancian, senior adviser at the Center for Strategic & International Studies, told us in an email.
Ball wrote on X that the U.S. “is the only participant in the war that is known to have Tomahawk missiles.”
In addition, Gen. Dan Caine, the chairman of the Joint Chiefs of Staff, said at a news conference on March 4 that the initial U.S. airstrikes were focused in the south of Iran, where the school bombing occurred. Israel “predominantly” targeted air defense systems in Iran’s “northern flank,” he said.
“An Israeli military official said the military was looking into the school incident but wasn’t aware of an Israeli strike in that area” with the school, the Wall Street Journal reported.
When asked about the school bombing, a U.S. Central Command spokesman, Capt. Tim Hawkins, told reporters that “it would be inappropriate to comment given the incident is under investigation.”
Meanwhile, Reuters, citing two unnamed U.S. officials, reported on March 5 that “U.S. military investigators believe it is likely that U.S. forces were responsible for an apparent strike on an Iranian girls’ school.” U.S. officials requesting anonymity to speak about the preliminary findings told the Associated Press, CBS News and the Wall Street Journal the same thing.
CBS News said “[t]he preliminary U.S. assessment suggests that the United States is ‘likely’ responsible for the deadly attack but did not intentionally target the school and may have hit it in error, possibly due to the use of dated intelligence which wrongly identified the area as still part of an Iranian military installation.”
In response to early reports about the probe, a White House spokeswoman, Anna Kelly, issued a statement to reporters saying that the “investigation is ongoing” and has reached “no conclusions at this time.” She called it “both irresponsible and false for anyone to claim otherwise.”
Reuters said in its reporting that the officials it spoke with “did not rule out the possibility that new evidence could emerge that absolves the US of responsibility.”
Even with satellite images and video of the airstrikes, remnants of the missile would need to be examined to more definitively determine culpability, N.R. Jenzen-Jones, an arms and munitions intelligence specialist who directs the Armament Research Services, told the newswire.
Complicating matters, the AP said, is the fact that “[n]o independent agency has reached the site during the war to investigate.”
At the March 9 press conference, Trump was asked why he is the only person in the U.S. government claiming that Iran was responsible for the bombing of the school. He replied: “Because I just don’t know enough about it. I think it’s something that I was told is under investigation, but Tomahawks are used by others, as you know. Numerous other nations have tomahawks. They buy them from us.”
But Cancian told us that the only countries other than the U.S. using Tomahawks are the United Kingdom, Australia, Japan and the Netherlands.
The U.K. and Australia have previously purchased the missiles, according to their own defense departments. The U.S. State Department approved selling the weapons to Japan and the Netherlands, in 2023 and 2025, respectively.
Those four countries are not involved in the U.S-Israeli conflict with Iran.
Ultimately, once the investigation is complete, “whatever the report shows, I’m willing to live with that report,” Trump said.
During a speech in South Carolina on Feb. 27, Joe Biden touted his record as president while criticizing his successor, President Donald Trump. But during his remarks, Biden made a number of false, misleading or exaggerated claims.
Biden was in South Carolina to celebrate winning the state’s Democratic presidential primary six years earlier. Biden’s win there helped propel him to become the Democratic nominee for president in 2020.
When Biden compared his jobs record with Trump’s, he exaggerated the figures.
“In fact, [in] just my last year as president of the United States in 2024, we created — just the last year — 2.2 million additional jobs,” he said. “You know how many jobs Trump’s created in his first year as president? 185,000 jobs total. That’s it.”
However, the most recent data from the Bureau of Labor Statistics show that total employment increased by a little more than 1.2 million between January 2024 and January 2025, which covers Biden’s last full year in office. (He left office on Jan. 20, 2025.)
Meanwhile, in the first full year of Trump’s second term, employment increased by 359,000, from January 2025 to January 2026.
For his speech, Biden may have relied on outdated data, or data covering a different period. We reached out to his office about his claims, but we didn’t receive a response.
BLS did report in January 2025 that total employment had increased by 2.2 million in 2024. That covers most of Biden’s final year as president. But that report came out before the BLS made annual data revisions for the 12 months ending in March 2025 that lowered its estimates of the increase in employment during Biden’s time in office. The final revisions were made on Feb. 11.
The latest BLS data also show that total employment in 2025 increased by 181,000, when measured from December 2024 to December 2025. That’s close to the 185,000 figure that Biden used for Trump. But Trump took office on Jan. 20, 2025, and BLS bases its job figures on a monthly survey of households covering the week that contains the 12th day of the month. That means the January 2025 job numbers were under Biden.
We got an increase of 359,000 for Trump by measuring from January 2025 to January 2026, which more closely aligns with the period covering his first full year back in office.
We would also note that the employment for January 2026 is preliminary and subject to be revised. Also, as we’ve said before, presidents shouldn’t receive all the credit, or the blame, for employment figures on their watch.
Biden also claimed that “the economy grew with record growth” during his presidency. We found no basis for that statement.
Real gross domestic product (meaning it has been adjusted for inflation) grew by 34.9% in the third quarter of 2020 and by 18.9% in all of 1942, which are the quarterly and annual economic growth records, according to Bureau of Economic Analysis estimates. The highest quarterly GDP growth under Biden was 7% in the second and fourth quarters of 2021, when the economy was rebounding from the COVID-19 pandemic, and the highest annual GDP growth during his administration was 6.2% that same year.
Average annual growth during Biden’s four years was 3.6%. That was still lower than the almost 4.5% average during Bill Clinton’s second term, and the average of nearly 5.2% during Lyndon B. Johnson’s full four-year term. There was even average annual growth of about 15.4% in Franklin D. Roosevelt’s third term, during World War II.
Taking out the bounce-back year after GDP plunged as a result of the pandemic, economic growth in the last three years of Biden’s presidency was about 2.7% annually, which is close to the yearly average of about 2.8% annual growth over the last 50 years.
Biden later turned to the subject of immigration, saying, “The day I left office, border crossings in the United States were lower than the day that I entered an office inherited from Trump.” That’s accurate, but misleading.
Border Patrol made 47,320 apprehensions of people illegally crossing the U.S. border with Mexico in December 2024, Biden’s last full month in office. Then apprehensions at the southern border declined further to 29,105 in January 2025, and Biden left office a little more than halfway through that month.
Those figures were down from 71,141 apprehensions by Border Patrol in December 2020, the last full month of Trump’s first term, and 75,316 in January 2021, when Trump exited the White House.
But in our story “Biden’s Final Numbers,” which looks at various statistical measures during his presidency, we wrote: “Illegal border crossings, as measured by apprehensions at the southwest border, were 107% higher in Biden’s final year in office compared with the last full year before he was sworn in, according to data from U.S. Customs and Border Protection.” We also said “that snapshot undersells the surge in illegal immigration during Biden’s four years in office, because apprehensions dropped dramatically in the second half of 2024 after Biden initiated some emergency policies to curb illegal border crossings.
“Before then, the U.S. was experiencing historically high illegal immigration,” we reported.
We also pointed out that apprehensions were only part of the picture, since the number of people seeking asylum at legal ports of entry remained high under Biden, as his administration began accepting CBP One mobile app applications that allowed immigrants to request asylum or parole and be screened for entry to the U.S. Plus there was an additional surge in immigrants coming to the U.S. via newly created legal methods, such as noncitizens granted parole, which allows them to temporarily live in the U.S. for “urgent humanitarian or significant public benefit reasons.” Biden offered parole to immigrants from countries such as Cuba, Haiti, Nicaragua and Venezuela. (Trump has largely halted those humanitarian programs through executive orders.)
While Biden suggested that the increase in migration earlier in his presidency was due to the pandemic, Julia Gelatt, associate director of the U.S. immigration policy program at the Migration Policy Institute, previously told us that there were several reasons for the surge.
“There were many different drivers in the growth of the unauthorized immigrant population during the Biden presidency: strong labor demand in the U.S. as the country rebounded from the COVID-19 recession, and push factors such as authoritarian governments in Cuba, Nicaragua, and Venezuela and intense gang violence and extortion in countries like Haiti and Ecuador,” she said. “It’s also possible that some people moved in order to take advantage of new pathways created by the Biden administration.”
Correction, March 12: Border Patrol apprehensions for December 2020 and January 2021 were 71,141 and 75,316, respectively, as confirmed by CBP. We had slightly different figures in our original article.
