Trump Releases Election Security Papers

How Secure Is Your Vote?

Government security testers broke into election networks in hours, not months. The same declassified files that expose the weaknesses also spell out the repairs and there is still time to make them before November.

A plain-language briefing built entirely from the records released by the White House Government Transparency Task Force.

When a team of federal cybersecurity specialists set out to test the computer networks that American communities use to run their elections, they did not need weeks of effort or insider access. In case after case, they were inside with the running of the network before most offices would have noticed anything was wrong. That finding, buried in a technical report only recently made public, is the reason this article exists. It is also, oddly, a hopeful story: because the very documents that catalog the weaknesses also list, in plain terms, exactly how to fix them.

In July 2026, the White House Government Transparency Task Force declassified and published a set of intelligence and law-enforcement records dealing with the security of U.S. elections. Much of the coverage that followed focused on foreign adversaries and political blame. This piece does something narrower and more useful: it reads the files for what they say about how election systems can be broken into, and what the same files recommend to keep that from happening. Every claim below is tied to the document it comes from, quoted word for word, so you can check it yourself. Where the original documents blacked out text, that is shown as [redacted].

The break-in that took hours, not months: The centerpiece of the release is a report by the Cybersecurity and Infrastructure Security Agency CISA, the federal agency responsible for helping protect election infrastructure. Between roughly 2019 and 2024, at the invitation of the state and local offices that own the systems, CISA’s specialists ran the kind of tests a real attacker would: they probed networks, reviewed the software’s source code, and attempted, with permission, to break in. What they found about the networks surrounding election systems is stated without euphemism.

“CISA assessors gained full network control within hours or days, demonstrating that many SLTT partners remain soft targets incapable of stopping even moderately skilled adversaries.”(1)

“SLTT” is government shorthand for state, local, tribal, and territorial the thousands of offices that actually administer elections. The report’s point is that the danger is rarely the voting machine sitting in the polling place. It is the ordinary office network around it: the email accounts, the staff laptops, the shared logins. Attackers who get a toehold there can move sideways toward the systems that matter. “election infrastructure is often accessible from general enterprise networks, creating opportunities for lateral movement by adversaries who compromise email systems, user workstations, or other information technology (IT) assets.” (1)

Many offices assume their election systems are “air-gapped” physically walled off from the internet and everything else. The testers found that assumption is often a comforting fiction, undone by a vendor’s remote-support connection or a forgotten network path. “Overreliance on “airgap” assumptions that do not reflect actual connectivity (e.g., vendor support tunnels, unmonitored remote management tools, or indirect network paths).” (1)

The weak point is usually not the ballot box it is the office next to it. An election system can be built to a high standard and still be reachable through the clerk’s email inbox or a vendor’s maintenance line. That is why the fix is rarely “buy better voting machines” and almost always “wall the election systems off from everyday office computing and lock the doors that connect the two.”

The repair the files prescribe: The registration-security companion document is blunt about the remedy hard separation, with a specific warning not to let election results travel across the ordinary business network: “Implement and enforce network segmentation. Proper network segmentation is an effective security mechanism to prevent an intruder from propagating exploits or moving laterally within an internal network. This includes not transferring election results on the business network.” (2)

The same document is equally direct about the doors attackers walk through most often weak or shared passwords and the single control that closes most of them: “Enable Multifactor Authentication on all accounts.”(2)

Machines that can’t be double-checked: Some of the exposure is in the voting equipment itself. The CISA report singles out a specific design in wide use a ballot-marking device that prints your choices as a barcode, which the machine, not you, reads back.”in 2020, Image Cast X Ballot Marking Devices printed voters’ completed ballots on paper but encoded their selections in a barcode that voters had no way to verify. A researcher showed that hackers could change the votes encoded in the barcode, without even having physical access to the machines.”(1)

The problem is not that this happened in any particular election the report does not claim that but that the design makes the counted vote something a voter cannot personally confirm. The fix the report gives is refreshingly low-tech: ballots people can actually read, and hand audits afterward to prove the machines counted them correctly. “Use human-readable paper ballots.” (1) “Conduct post-election manual audits of paper ballots to confirm that voting systems function as intended and to identify errors prior to certification of results.”(1)

There is a second, structural weakness behind the equipment: the rules meant to keep voting systems trustworthy can also keep them from being fixed. Certification regimes often freeze the software in place for months before an election, so known flaws simply sit there. “some government election systems certification regimes require that no patches be applied for months before an election.” (1) “In practice, these constraints produce environments where known, documented vulnerabilities persist for months or years on production election systems, increasing exposure to opportunistic and targeted threats.” (1)

The report’s answer is to change the rule itself, so a security fix no longer forces a choice between “patched” and “certified”: “Harmonize relevant patch management and certification rules for voting systems and associated IT infrastructure to allow cybersecurity changes to be made in real-time, without impacting certification.” (1)

The data that never goes stale: If the voting machines are the part of the system voters can see, the voter-registration databases are the part they can’t and, according to the files, the part adversaries want most. A dedicated report in the release, “Recognizing and Addressing Threats to Statewide Voter Registration Databases,” lays out how relentlessly these systems have been targeted.There were 50 states where hackers have attempted to breach voter registration systems. Voter Registration Database Threats totaled 20+states with confirmed successful breaches.Voter Registration Database Threats200M+voter records reported compromised by the PRCWHTF Task Force statement “Hackers have attempted to breach voter registration systems in all 50 states, with confirmed successes in at least 20 of them.” (2)

The report’s most important argument is about why this matters, and it pushes back on the idea that a stolen voter file is merely embarrassing. The danger, it says, is what the data lets someone do later and it never expires. “The impact of the breaches is not limited to “undermining confidence” or “spreading false claims” but the data itself could be used months or years after the breach to alter voter registration information or request absentee ballots. The real threat is what can be done with the stolen data.” (2) “Data obtained in a breach from 2021, for example, could be used to request a ballot for an election in 2028 because the data does not get stale.” (2)

It is critical to understand that a stolen voter file is a slow-acting weapon. Unlike a stolen credit card, which is canceled within days, the facts that verify a voter name, birth date, address, driver’s-license and Social Security fragments do not change. Data taken years ago can still be used to request someone’s ballot or quietly change their registration today. That is why the defense the files emphasize is not just “keep attackers out,” but “make sure that even after a breach, every change to the rolls is logged, noticed, and reversible.”

The repairs that the files prescribe. Against a threat that lingers for years, the report’s counter is resilience: frequent offline backups that let an office restore altered or deleted records, and fast repair of known holes. “States should schedule frequent, routine back-ups of the voter registration database files and store them securely offline.” (2) “this requires critical vulnerabilities to be re-mediated within 15 calendar days of initial detection.” (2)

The everyday attack surface email and public websites get the same practical treatment, down to blocking the web addresses ransomware relies on and moving official sites onto authenticated .gov domains. “MDBR can block the vast majority of ransomware infections just by preventing the initial outreach to a ransomware delivery domain.” (2) “For jurisdictions that utilize electronic pollbooks, producing paper pollbooks as back-ups will limit the potential for disruptions on Election Day.” (2) The reason this database chapter matters so much is spelled out elsewhere in the release. The Task Force’s public statement ties the domestic weakness to a foreign campaign against the rolls: “The declassified intelligence reveals that voter registration rolls from at least 18 states (not all identified by name) have been compromised by the People’s Republic of China (PRC). Additional intelligence records reveal that more than 200 million voter records were also compromised by the PRC, without state-specific affiliations.” (5)

Whatever one makes of the geopolitics, the practical lesson for an election office is the one the report already drew: We should assume the data is out there and build defenses that limit what a copy of it can do.

Fraud at the front door: Not every threat in the files is a cyberattack. Some of it walks in through the ordinary paperwork of signing people up to vote. One set of documents traces an FBI investigation into a voter-registration drive in Muskegon, Michigan.”The FBI Detroit Field Office (FBI Detroit) was informed that the Michigan State Police (MSP), Michigan Attorney General (MAG), and Muskegon Police Department (MCPD) were investigating allegations of fraudulent voter registrations being submitted to the Muskegon City Clerk’s Office (MCCO).”(3) “MCPD estimated [redacted] and [redacted] had delivered between 8,000 to 10,000 registrations.”(3)

The details that matter most for election security is not the size of the pile it is where the fraud was caught. It was stopped at intake, by the city clerk’s own review of the applications, before it could become anything else. “Since on or about October 5, 2020, MCCO has received six packages, marked [redacted] via U.S. Mail containing voter registration applications. MCCO determined some of the applications were fraudulent based on a number of factors.” (3)

That is the lesson in miniature: careful screening at the point where applications enter the system is the control that works. A related concern in the release is the presence of noncitizens on voter rolls. A Department of Homeland Security summary states its findings in capital letters:

“OVER 250,000 NON-CITIZENS ARE ILLEGALLY REGISTERED TO VOTE IN JUST THE FOUR STATES FOR WHICH PUBLIC DATA FILES HAVE BEEN REVIEWED.” (4)

The same document points to a remedy it says other states already use a federal matching service that flags registrants who are not eligible: “DHS has enabled those states to identify over 400,000 deceased registrants and over 28,000 non-citizens who illegally registered to vote.”(4)

What can be done before November: The striking thing about this collection is how little of it is a mystery. The weaknesses are documented and so are the fixes, most of them unglamorous, well understood, and achievable inside a single budget cycle. Pulled together, the record points to a short list of moves any election office can weigh now.The files describe risk soft networks, unpatched machines, breached and lingering data, fraud attempts at intake not proof that any result was changed. That distinction matters, and it cuts toward action rather than alarm: the very same documents that name the vulnerabilities also name the repairs, and the repairs are ordinary security hygiene, not moonshots.

What to ask your election officials

1. Is multi-factor authentication on everywhere? Every account, no shared logins the single highest-value control in the files.
2. Are election systems walled off from office email and business networks? Including a promise that results never travel across the business network.
3. Are there human-readable paper ballots and post-election hand audits? So the count can be checked independently of the machines.
4. Are voter-registration databases backed up offline, with at least 30 days of recovery points? So altered or deleted records can be restored.
5. Are paper pollbook backups ready for any place using electronic pollbooks on Election Day?
6. Is registration intake screened rigorously, with every change to the rolls logged and reversible, and eligibility checked against federal matching services?

What the files do and don’t establish: Honesty about the limits of these documents is part of taking them seriously. The CISA report and the registration-threats report are detailed, internally consistent technical assessments, and their findings and fixes are the backbone of this article. Several of the more dramatic figures, however, are assertions by the releasing agencies rather than independently audited results: the claim that “more than 200 million voter records” were compromised, and the DHS count of “over 250,000 non-citizens” registered are stated here as those documents state them, attributed to the source, not confirmed by outside review.

It is equally important to be clear about what the files do not say. They describe vulnerabilities, intrusion attempts, breaches, and fraud attempts the potential for an election system to be compromised. They do not, in the material reviewed here, demonstrate that any election’s outcome was changed, and this article makes no such claim. Finally, several of the underlying records are scanned and heavily redacted; quotations from those documents were checked against the original page images so the wording is exact, but blacked-out passages (shown as [redacted]) necessarily limit how completely they can be read. None of this diminishes the core, well-supported findings: The security gaps are real, and the remedies are known and available before November.

Source documents: Every quotation above is verbatim from the following files, released by the White House Government Transparency Task Force. Bracketed “[redacted]” marks text withheld in the original.

1. CISA Election Report – FINAL.pdf Cybersecurity and Infrastructure Security Agency, “Election Report” (findings from hands-on testing of election systems, c. 2019–2024). Cited 9×
2. Voter Registration Database Threats – FINAL.pdf “Recognizing and Addressing Threats to Statewide Voter Registration Databases,” July 2026. Cited 8×
3. 0000001 – RELEASE MARKED.pdf FBI Detroit “Opening EC,” case 56D-DE-3407960 (Muskegon, Michigan registration matter). Cited 3×
4. Alien Voter Registration Summary.pdf_Redacted.pdf  Department of Homeland Security, “Preventing Alien Voting” summary. Cited 2×
5. WHTF Government Transparency – States Statement FINAL.pdf  White House Government Transparency Task Force public statement, July 13, 2026. Cited 1×

The vulnerabilities in these files are not a verdict on any past election. They are a to-do list for the next one and the clock runs to November. American voters need and expect a secure voting system that cannot be compromised.

“We Get the Government We Deserve”