The SAVE Act: What They're Selling vs. What's in the Bill

CBS/YouGov · March 16–19, 2026 · 2,500 adults surveyed

UMass Amherst/YouGov · March 20–25, 2026 · 1,000 adults surveyed

Support by race: 56% white · 43% Latino · 39% Black. The communities most likely to lack documentary proof of citizenship are the least likely to support requiring it.

The poll directors stated: "Americans may support the principle of citizenship verification in the abstract while remaining genuinely uncomfortable with who gets caught in the net."

The Priority Test

The 80–90% figures measure support for photo ID requirements — not the SAVE Act. These are different questions about different policies.

CBS News/YouGov asked about the SAVE Act by name: 28% favor, 31% oppose, 41% not sure. Only 16% of Republicans say they know "a lot" about the bill. When voters learn what's actually in it — the documentation requirements, the incentive structure, the implementation timeline — support collapses.

The same CBS/YouGov poll found 80% support photo ID. That 52-point gap between the concept and this bill is the story. Provisions poll higher than the bill because the details change minds.

Existing registration protects you only if nothing in your life changes. Anyone who moves, changes their name, or updates their registration must re-register with full documentary proof under the SAVE Act. That covers 69 million women with name mismatches between their birth certificate and current ID, military families who relocate frequently, college students, and anyone who changes address.

The bill also requires states to take "affirmative steps on an ongoing basis" to cross-reference voter rolls against federal databases. Flagged registrants must prove citizenship or be removed — and the bill contains no requirement that states notify voters before removal. You could be purged and not find out until Election Day.

Virginia's 2025 purge documented the real-world result: 94% of those removed were U.S. citizens.

It's different because voter ID and documentary proof of citizenship are different requirements. The 36 states that require voter ID accept driver's licenses — the most common form of government-issued ID in America. The SAVE Act does not accept driver's licenses for registration because a driver's license does not prove citizenship. Non-citizens on visas and green cards have always been eligible for state driver's licenses.

The argument that "36 states require voter ID" without distinguishing what those laws actually require conflates a requirement Americans already meet (showing a DL at the polls) with one most Americans have never faced (producing a passport or birth certificate to register). These are not the same policy. They are not even close.

Of the 36 states with voter ID laws, only Ohio has a standard as restrictive as the SAVE Act. The bill would override the existing, less restrictive voter ID frameworks that 35 states chose for themselves — imposing a federal mandate stricter than almost every state's own law.

The same political coalition that opposes federal election standards when proposed by Democrats is imposing them here. The Freedom to Vote Act proposed national voter ID with broader accepted documents. Not a single Republican voted for it.

The SAVE Act accepts a REAL ID only if it "indicates the applicant is a citizen of the United States." Standard REAL IDs do not do this. They verify lawful presence — which includes green card holders, visa holders, and other non-citizens. A REAL ID star means the card meets federal security standards. It does not mean the holder is a citizen.

Only 5 states — Michigan, Minnesota, New York, Vermont, and Washington — issue Enhanced Driver's Licenses that explicitly indicate U.S. citizenship. These are distinct from standard REAL IDs: they carry a U.S. flag marking, cost more, and are designed for border crossing. The TSA confirms that EDLs provide "proof of identity and U.S. citizenship" — standard REAL IDs do not.

For voters in the other 45 states, a REAL ID does not satisfy the SAVE Act. They still need a passport, birth certificate, or naturalization certificate. The "just use your REAL ID" argument sounds reasonable until you read the bill's actual text — which most voters, and apparently some legislators, have not.

Timothy Scouton pleaded guilty to letting 11 people skip registration paperwork — not 11 non-citizens voting. No citizenship issue was alleged at any point in the case. The SAVE Act governs citizenship documentation at registration. Scouton bypassed the registration process entirely — directing staff to skip forms altogether.

An official willing to circumvent one requirement will circumvent all of them. The SAVE Act would not have stopped this. The fix for official misconduct is prosecution — which is exactly what happened. The Minnesota Secretary of State said the case was caught immediately and that the office is unaware of any other case like it.

The system worked. That's the receipt, not the argument against it.

This is the ecological fallacy — two unrelated statistics juxtaposed to imply a connection that doesn't exist. California jury pools are drawn from DMV lists AND voter rolls. Non-citizens can obtain California driver's licenses under state law, so they appear in jury pools via DMV data, not voter registration.

California's Deputy Secretary of State called the claim "false and inaccurate." County-level investigations found no overlap: in Nevada County, 97 potential jurors were excused for non-citizenship — and a search of their names on the voter registration list found zero matches. In Sierra County, 3 were dismissed for non-citizenship; none were registered voters.

The claim has been debunked by PolitiFact, Snopes, and Lead Stories. It has circulated in various forms since at least 2019. No evidence has ever been produced that those 449K were registered voters.

A recurring pattern on social media: list a large number — mail-in ballots cast, registered voters, ballot requests — next to a senator's name or a party label, and let the reader assume the volume itself is evidence of fraud. No claim is made. No anomaly is identified. The numbers are just presented.

This is proof by juxtaposition — a rhetorical device where statistics are placed side by side so the reader's brain fills in a connection the poster never had to defend. It works because a big number feels alarming, and debunking it requires explaining systems the original post deliberately skipped.

Mail-in ballots aren't unverified. Every state that conducts mail voting uses signature matching, unique barcodes tied to individual registrations, and ballot tracking systems. "No photo ID required" is not the same as "no verification." It means the state uses a different verification method — one that in most cases was enacted by bipartisan or Republican-majority legislatures.

Voter rolls are designed to lag behind reality. The National Voter Registration Act restricts how aggressively states can remove names. People die, move, or become ineligible — their registrations persist until the state completes a maintenance cycle. Outdated registrations are not votes. They are administrative lag built into the system by federal law.

Census citizenship estimates are estimates. The American Community Survey's margin of error can reach ±50,000 or more in smaller states. A "gap" of a few thousand between registered voters and estimated eligible citizens may not be a gap at all.

The actual audits find almost nothing. Utah audited every one of its 2 million+ registrations specifically looking for noncitizen voting. Result: 1 noncitizen registration, 0 votes cast. Georgia audited 8.2 million: 20 noncitizens found, 9 had ever voted.

The question to ask every time: Did the person posting this identify a specific anomaly, or did they list large numbers and let you assume the rest? A subtraction problem is not an audit. A big number next to a name is not evidence. If the volume of legal mail-in voting is the concern, the argument is against mail voting itself — not against the voters who used it.

No. This is a conflation of active registrations with inactive records — and it's becoming a coordinated messaging pattern.

The technique works like this: take the total number of entries in a state's voter database — which includes active voters, deceased voters, people who moved out of state, and people convicted of felonies — and compare it to the adult population. The resulting number is always alarming. It's also meaningless.

Wisconsin is the most prominent current example. Social media posts claim 8.3 million "registered voters" in a state with 4.8 million adults. The actual number of active registered voters is 3,601,785 (Wisconsin Elections Commission, February 2026). The 8.3 million figure includes 4.6 million inactive records that state law requires be maintained.

The inactive list is a fraud prevention tool, not a vulnerability. When a voter is flagged as deceased, moved, or ineligible, their record moves to inactive — but it isn't deleted. That's by design. Keeping the record means clerks can catch it if someone tries to register under that name again. Deleting the record would remove the paper trail and make fraud easier, not harder. The Wisconsin Elections Commission chair confirmed this directly: inactive records are kept indefinitely so that if someone tries to register using a dead voter's identity, clerks can flag the application because the prior record still exists.

Pollbooks used on Election Day contain only active voters. The broader database retains inactive records for exactly the fraud-prevention purpose described above. An inactive voter cannot walk into a polling place and vote — they would need to re-register and clear the verification process again.

The claim that someone "could flip 50,000 inactive voters to active in seconds" is conspiracy theorizing. It would require unauthorized access to a secured state database, would create an audit trail, would produce registrations with no matching voter appearing at polling places, and would be immediately detectable in post-election canvassing. Every state with electronic voter registration maintains access logs.

The pattern is worth noticing: the same accounts presenting inactive records as evidence of fraud are simultaneously opposing ERIC — the bipartisan interstate system that helps states maintain accurate rolls by flagging duplicates, deceased voters, and cross-state moves. You can't claim to care about voter roll accuracy while dismantling the tools that maintain it.

When someone presents a number that includes inactive records without distinguishing them from active voters, ask: are they confused about how voter databases work, or are they counting on you being confused?

Yes — and existing law already does that. The problem with the SAVE Act isn't that it holds officials accountable. It's that it uses strict liability instead of a negligence standard, which means it punishes honest errors the same as intentional violations.

Under a negligence standard — the legal framework used in virtually every other professional liability context — an official who follows written procedures, acts in good faith, and has no actual knowledge of ineligibility is protected. An official who knowingly registers an ineligible voter faces full criminal penalties. The distinction tracks intent.

Under the SAVE Act's strict liability standard, that distinction disappears. An official who follows every procedure correctly, makes a reasonable judgment call on an ambiguous case, and turns out to be wrong faces the same exposure as one who deliberately commits fraud — up to five years federal prison plus private civil lawsuits from any individual, with no merit requirement.

Strict liability is appropriate in narrow contexts where the harm is so severe and the actor's control so complete that intent is irrelevant — environmental contamination, product liability. It is not appropriate where officials are making high-volume, time-pressured judgment calls on ambiguous documentation with no clear guidance, no funding, and no training.

The SAVE Act's choice of strict liability over negligence is not incidental. A negligence standard would hold bad actors accountable while protecting good-faith officials. Strict liability holds everyone accountable for every error — which produces one predictable outcome: officials stop making judgment calls and default to rejection in every uncertain case.

The escape hatch exists on paper. The incentive structure guarantees it won't be used.

Under the SAVE Act, an election official who accepts a registration without full documentary proof faces up to five years in federal prison plus private civil lawsuits — filed by any individual, with no merit requirement and no cap on legal costs. The official who rejects an eligible citizen faces nothing. Zero penalty. Zero liability.

Before EAC guidance exists — and the bill takes effect immediately, with guidance arriving ten days later while primaries are already underway — officials have no framework at all. When guidance does arrive, if it is vague or silent on a specific document combination (virtually guaranteed given the complexity of citizenship documentation across 330 million people), officials face the same calculation: prison and lawsuits for accepting, nothing for rejecting. The default will be rejection every time.

Decades of behavioral economics research establish why. Kahneman and Tversky's foundational work on loss aversion demonstrates that people weight potential losses roughly twice as heavily as equivalent gains. When the asymmetry is this extreme — prison and lawsuits on one side, nothing on the other — rational actors don't use the escape hatch. They default to rejection in every case of uncertainty.

This isn't speculation. Kansas implemented a similar documentary proof requirement and blocked 30,000+ eligible citizens while stopping 0.002% non-citizen registrations. Virginia conducted voter purges and 94% of those removed were U.S. citizens. New Hampshire implemented proof-of-citizenship requirements and turned away 244 eligible voters on day one. In each case, the incentive to reject produced exactly that outcome.

The closest comparable framework is I-9 employment verification, which also requires citizenship attestation. Federal law explicitly builds in a good faith safe harbor: employers who follow written procedures and act without actual knowledge of ineligibility are protected from penalty. The SAVE Act omits that standard entirely. That's not consistent with how citizenship attestation works elsewhere — it's the exception.

In theory. In practice, it creates an organized intimidation infrastructure with no merit threshold.

Any individual can file a civil lawsuit against an election official for accepting a registration — with no requirement that the plaintiff believes anything improper occurred, no requirement of harm, and no good faith threshold. The bill creates a cause of action that can be weaponized by anyone, for any reason, at any time an official uses the escape hatch. The official must retain legal counsel and mount a defense regardless of outcome.

The average election clerk earns approximately $32,000 per year. The average county clerk earns approximately $48,000. A single civil suit — even a completely meritless one filed by someone who has no belief that anything was wrong — requires legal representation, time, and professional exposure that can be financially ruinous for a local government employee at that salary level.

A well-funded outside group — a PAC, a nonprofit, or a single wealthy individual — could systematically monitor registration offices, identify officials who use the attestation process, and file coordinated civil suits against them. The suits don't need to succeed. The filing itself is the deterrent. We have already seen nine-figure sums flow into election-related political action committees in recent cycles. The same funding infrastructure that drives political advertising can drive coordinated litigation campaigns against individual county clerks.

This is not a hypothetical concern. It is the predictable application of an existing and well-documented political funding infrastructure to a new statutory tool that requires no merit, no harm, and no good faith to deploy.

The result isn't accountability for bad actors. It's targeted suppression of the one official behavior — using the escape hatch — that the bill nominally permits. An official who knowingly registers an ineligible voter should face consequences. An official making a good-faith judgment call on ambiguous documentation should not face a litigation campaign funded by outside groups with no stake in the individual case.

It is worth asking why the bill was structured this way. A genuine escape hatch — one intended to be used — would include a negligence standard, a safe harbor for officials following written procedures, and no private right of action for non-meritorious claims. EISA includes all three. The SAVE Act includes none of them. That's not an oversight.

Crawford upheld Indiana's voter ID law. It did not uphold documentary proof of citizenship, and the reasoning in Crawford actually undermines the case for the SAVE Act.

Justice Stevens's lead opinion turned on a specific finding: because Indiana provided free photo IDs, the burden on voters was "limited." The opinion states explicitly that the burden was acceptable because "Indiana's cards are free" and the "inconvenience of going to the Bureau of Motor Vehicles, gathering required documents, and posing for a photograph does not qualify as a substantial burden on most voters' right to vote."

The SAVE Act requires documents that are not free. A passport costs $130–$160. Replacement birth certificates cost $10–$60 depending on the state and take weeks to months to process. The bill provides zero funding for document acquisition. That's not the Crawford fact pattern — it's closer to the one Crawford distinguished itself from.

Harper v. Virginia Board of Elections (1966) struck down a $1.50 poll tax — the Court held that "to introduce wealth or payment of any fee as a measure of a voter's qualifications is to introduce invidious discrimination." The SAVE Act imposes costs 80 to 100 times that amount with no fee waiver and no funding mechanism.

Crawford also noted that the "severity of the somewhat heavier burden" on voters without ID was "mitigated" by Indiana's provisional ballot process. The SAVE Act's provisional ballot mechanism requires the voter to produce the same documentary proof after the fact — there is no alternative path to having the ballot counted without the documents the voter didn't have in the first place.

The gap between Crawford (free ID, provisional ballot safety valve) and the SAVE Act ($130+ documents, no funding, no meaningful provisional alternative) is precisely the constitutional exposure that Crawford's own reasoning creates.

Kansas tested this theory. 31,089 eligible citizens couldn't clear it — 12% of everyone who tried to register for the first time. A federal court struck the law down. The 10th Circuit affirmed.

Replacement birth certificates cost $10–$60 depending on the state and processing method. Standard processing takes 2 to 12 weeks. Passports cost $130–$160 and take 6–8 weeks. Many Americans born before modern record-keeping — particularly Black Americans born in the rural South, Native Americans born on reservations, and older citizens born at home — have birth certificates with errors, missing records, or no certificate at all. The University of Maryland found 3.8 million adult citizens have no documentary proof of citizenship whatsoever.

The bill provides zero funding for document acquisition, processing, or outreach. It mandates the requirement and delegates the cost entirely to individual voters.

Here is the comparison that clarifies the standard: to get a job in America, federal I-9 employment verification accepts a driver's license plus a Social Security card. The SAVE Act demands more to vote than the federal government requires to work. If a DL and Social Security card are sufficient to verify someone is authorized to work in the United States, the question is why they aren't sufficient to verify someone is authorized to vote.

New Hampshire implemented a similar requirement and 244 people were turned away during low-turnout local elections alone — including a U.S.-born veteran who had previously voted in the state. The "it's not hard" premise has been tested. The results are documented.

The SAVE Act overrides the voter ID laws of 35 of 36 states by imposing a documentation standard stricter than what those states chose for themselves. It mandates that states share unredacted voter rolls with a federal agency — DHS — with no restrictions on use. It dictates an EAC guidance timeline of 10 days and a roll submission deadline of 30 days. It prescribes which documents states must accept and which they cannot. It subjects state and local election officials to federal criminal penalties for how they administer state voter registration.

Whether you call that "federalizing elections" is a semantic question. What it does, functionally, is impose federal control over state election administration procedures — the same thing the bill's sponsors say they oppose when proposed by the other party.

Utah Lt. Governor Deidre Henderson — a Republican and the state's chief elections officer — made this point directly: "If we want a federal law mandating voter ID or 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. That's not what's happening with the SAVE America Act."

The Constitution gives states primary authority to administer elections. Congress has the power to override — Article I, Section 4. But it is worth noting who invokes that power and when. The Freedom to Vote Act proposed national voter ID standards with a broader accepted documents list. It received zero Republican votes. The SAVE Act proposes a stricter national standard with narrower accepted documents. It received near-unanimous Republican support. The principle is not "states should control elections." The principle is "federal standards are acceptable when they restrict access and unacceptable when they expand it."

No. Noncitizen voting in federal elections has been a federal crime since 1996, under the Illegal Immigration Reform and Immigrant Responsibility Act (IIRIRA). No federal or state Democratic lawmaker has proposed legalizing it. The claim that opposition to the SAVE America Act equals support for noncitizen voting is a non sequitur — it does not follow from the premise.

The documented opposition to the bill is specifically about implementation, not the goal. Critics — including Republican senators like Lisa Murkowski and Thom Tillis — have raised concerns about the bill blocking millions of eligible citizens from registering, not about allowing noncitizens to vote. The bipartisan Husted voter ID amendment, which stripped out the most problematic provisions, had Republican sponsorship. The disagreement is about method, not goal.

What the evidence shows: The Heritage Foundation's 20-year database of election fraud finds fewer than 100 confirmed cases of noncitizen voting nationally. Utah audited 2M+ voter records and found 1 noncitizen registration, 0 votes cast. Georgia found 9 noncitizen ballots out of 8.2M cast. These are the numbers from the most rigorous audits conducted by Republican-led states.

What is documented: The SAVE Act at Kansas's documented rate would block approximately 2.8 million eligible citizens from registering — a 1,000-to-1 ratio of citizens blocked to noncitizens stopped. Opposing that ratio is not the same as wanting noncitizens to vote.

This is a legitimate question that deserves a direct answer. States do share voter data with the Electronic Registration Information Center (ERIC) — a multi-state voter roll maintenance compact. The distinction between ERIC sharing and DHS sharing turns on three factors:

Purpose limitation. ERIC is a state-controlled compact with contractual restrictions on how data can be used — limited to voter roll maintenance. The DOJ/DHS requests contained no equivalent restrictions. That matters because a DOGE team member at the Social Security Administration signed a "Voter Data Agreement" with an outside advocacy group whose stated aim was to "find evidence of voter fraud and to overturn election results in certain states." That is documented in a DOJ court filing from January 2026. ERIC's governance structure makes that scenario impossible by design.

Oversight structure. ERIC is governed by member states with defined exit rights. DHS/DOGE has no equivalent accountability mechanism for voter data use. Three federal courts have already ruled states are not required to comply with DOJ demands for voter rolls.

The ERIC mailer problem is real but separate. Colorado sending voter registration mailers to 30,000 non-citizens is a documented ERIC design flaw — the EBU (eligible but unregistered) mailer requirement does not screen for citizenship before mailing. That is worth fixing. But the fix is reforming ERIC's mailer screening process, not handing unredacted voter rolls to DHS with no statutory use restrictions. EISA's database-first citizenship verification addresses the underlying gap without creating an unrestricted federal voter database.

Yes. It failed. Then we built something better. Then states left it. Now the SAVE Act proposes a federal version of what failed.

Crosscheck (2005–2018): Kansas ran the Interstate Voter Registration Crosscheck Program across 28 states. It matched voter records on first name, last name, and date of birth — essentially a Ctrl+F across state databases. The result: a 99%+ false positive rate. In 2017, the program flagged 7.2 million "potential duplicates" from 98 million records. Virginia purged 41,637 voters in a single year based on Crosscheck lists. Ada County, Idaho removed 766 voters — none were actual duplicates. The program disproportionately flagged voters of color: 1 in 6 Latino voters, 1 in 7 Asian Americans, 1 in 9 Black voters in participating states. Security was so poor that DHS intervened, and the database holding nearly 100 million voter records could be "hacked by a novice." The ACLU sued. Kansas settled. The program was killed.

ERIC (2012–present): The Electronic Registration Information Center was built specifically to fix Crosscheck's problems. Instead of crude name matching, ERIC uses voter data, DMV records, SSA death records, and USPS change-of-address data — producing rare false positives. It was founded by seven states, four of them Republican-led. It grew to 33 states and the District of Columbia. Even Jon Husted praised it as Ohio Secretary of State in 2016, calling it "a robust and secure system" that made it "easy to vote and hard to cheat." Then nine Republican-led states withdrew in 2022–2023, citing conspiracy theories. They left with no replacement. Election experts noted that the alternatives these states pursued looked "very similar" to Crosscheck.

The SAVE Act (2026): The bill requires states to submit unredacted voter rolls to DHS within 30 days for cross-referencing through the SAVE system. USCIS itself says the system "cannot reliably determine citizenship." When Travis County, Texas tested SAVE-flagged voters, 25% had already proved their citizenship. The bill includes no notification requirement before removing someone from the rolls — and criminalizes election officials who register someone without documentary proof, even if the voter is a citizen.

The country already built a better system. States left it. Now the SAVE Act proposes a federal version with less data, less accuracy, more power, and criminal penalties for the officials who administer it.

No. After two weeks of extended debate beginning March 17, the Senate recessed on March 28 without passing the bill. The Senate returned from recess on April 13 and has resumed floor attention on the bill, though no new votes have been scheduled as of April 16.

The bill needed 60 votes to overcome a filibuster. Republicans hold 53 seats. The closest they came was the Husted voter ID amendment on March 26, which failed 52-47 — every Democrat voted against it, and Sen. Lisa Murkowski (R-AK) voted against proceeding to the underlying bill on March 17. Cloture failed 49-41 on March 21. Sen. Thom Tillis (R-NC), a supporter of voter ID in principle, described the floor debate as having "lived up to my expectations for a nonoutcome."

Republicans are now pursuing a second path: including SAVE Act provisions in a budget reconciliation bill, which requires only a simple majority. Senate Budget Committee Chairman Lindsey Graham announced plans to draft such a bill. However, the bill's own Senate sponsor, Mike Lee, called reconciliation "essentially impossible" for the SAVE Act because the Byrd Rule bars provisions whose budgetary effect is "merely incidental" to their policy purpose. The House Freedom Caucus posted that Senate Republicans "claiming they will pass the SAVE America Act via budget reconciliation are gaslighting you." Sen. Rick Scott said he doesn't "see any way that any part of the SAVE America Act with any teeth gets included in a reconciliation package."

President Trump has repeatedly demanded the Senate "TERMINATE THE FILIBUSTER" to pass the bill. Thune has declined.

Yes. As of April 1, 2026, at least three states have enacted or are enacting their own proof-of-citizenship laws modeled on the SAVE Act:

Florida: Governor DeSantis signed a proof-of-citizenship bill into law on April 1, 2026. The law requires election officials to verify citizenship through state motor vehicle records and mandates documentary proof from voters whose citizenship cannot be confirmed. The main provisions do not take effect until 2027 — after the midterms. Florida found 198 likely noncitizens out of 13 million+ registered voters in 2025.

South Dakota and Utah: Both enacted proof-of-citizenship legislation that takes effect before the 2026 midterms. Both would adopt Arizona-style bifurcated systems where voters without proof of citizenship can only vote in federal elections. (This sounds backwards, but it follows from a 2013 Supreme Court ruling: federal law prohibits states from rejecting the federal registration form, which doesn't require DPOC — so states must let those voters participate in federal races. States can still require DPOC for their own elections.)

Advancing elsewhere: Mississippi has passed bills in both chambers (awaiting reconciliation). Iowa's Senate passed a version. Kansas's House passed one. Fifteen states total have introduced documentary proof of citizenship legislation this session, according to the Voting Rights Lab. New voter ID restrictions have advanced in at least one chamber in Arizona, Idaho, Iowa, Kentucky, New Hampshire, Oklahoma, Utah, and West Virginia.

This is the strategic landscape: even if the federal bill fails in the Senate, the provisions are being enacted state by state. Each state-level implementation will generate its own data on citizen impact — and that data will either validate or undermine the case for the federal version.

On March 31, 2026, with the SAVE Act stalled in the Senate, President Trump signed an executive order titled "Ensuring Citizenship Verification and Integrity in Federal Elections." The order attempts to accomplish by executive action what the bill could not achieve legislatively.

What the order does: Directs DHS and SSA to compile a "State Citizenship List" of confirmed U.S. citizens eligible to vote in each state. Instructs USPS to send mail ballots only to voters on an approved list. Threatens to cut federal funding to states that don't comply. Grants DOGE and DHS access to state voter files. Directs the Attorney General to prioritize prosecution of election officials who distribute ballots to ineligible voters.

Legal status: Three federal courts have already blocked the order's requirement that Americans show a passport or citizenship document to register using the federal form, ruling the president lacks authority to unilaterally alter election procedures. Multiple lawsuits have been filed. Election law expert Rick Hasen (UCLA) called the order "likely unconstitutional" and said implementation for 2026 is "virtually impossible." The DOJ has sued 29 states for refusing to hand over voter files; three courts have dismissed those suits.

The SAVE system problem applies here too. The order relies on the same DHS SAVE database that USCIS has acknowledged "cannot make confident determination of citizenship" in many cases. The same accuracy problems documented in the explainer — Travis County's 25% false-flag rate, persistent errors with naturalized citizens — apply to any citizenship list compiled from this data.

One detail: Trump himself voted by mail in Florida's special election earlier in March 2026 — weeks before signing an order restricting mail voting for everyone else.

Multiple legislative vehicles are being discussed. Rep. Anna Paulina Luna has threatened to "kill FISA unless the SAVE America Act is attached to it." FISA Section 702 — the surveillance authority — expires April 20, 2026. House hardliners have also proposed attaching it to DHS funding bills.

In the Senate, the reconciliation path announced by Budget Committee Chairman Lindsey Graham would attempt to include SAVE Act provisions in a filibuster-proof budget bill. But as documented in the Senate status FAQ above, the bill's own sponsor Mike Lee and the House Freedom Caucus have said this path is unviable due to the Byrd Rule.

Using the SAVE Act as a "poison pill" — attaching it to must-pass legislation to force Democrats to either accept it or block critical national security or government funding bills — is a documented legislative tactic. It prioritizes the political value of the vote over the policy outcome. Whether that strategy serves election integrity or election-year messaging is a question voters can answer for themselves.

Both chambers must pass identical text before a bill goes to the president. The House passed its version of the SAVE America Act on February 11, 2026. If the Senate passes the same text without changes, it goes directly to the president's desk.

If the Senate passes an amended version — even with minor changes — the two chambers must reconcile the differences. That typically happens one of two ways:

Conference committee: House and Senate leaders appoint members to negotiate a single unified text. Both chambers then vote on the conference report. No further amendments are allowed — it's an up-or-down vote. This is the traditional path for major legislation with substantive differences.

Ping-pong: One chamber passes an amended version and sends it back. The other chamber either accepts it or amends it again and sends it back. This continues until both chambers agree on identical text. This is faster but can stall if neither side concedes.

Either path means additional votes in both chambers — and additional opportunities for the bill to fail, be amended further, or be delayed past the midterms. Every amendment that changes the text restarts the clock. That's why Senate leadership has been careful about which amendments to allow: each change that passes creates a new version the House must also approve.

As of April 16, 2026, the Senate has not passed any version. Thune's procedural amendments have not substantively altered the House-passed text. But if the bill advances through reconciliation or with floor amendments, the conference process becomes the next bottleneck.