Election Report (CISA)

From ~2019 to 2024, CISA conducted technical and operational activities to evaluate certain US election systems, all upon request of system owners/operators, including software examination, penetration testing of SLTT (state, local, tribal, and territorial) networks, and incident response for election-system intrusions (p.1).

CISA notified the owner/operator in every case it identified a vulnerability in a product or network and encouraged mitigation (p.1).

Structural constraints in the certification ecosystem limit vendors' ability to patch quickly; some certification regimes require that no patches be applied for months before an election, so systems are deployed with known, unpatched issues (p.1).

SLTT election-office IT networks frequently lack cybersecurity hygiene; election infrastructure is often accessible from general enterprise networks, enabling lateral movement by adversaries who compromise email, workstations, or other IT assets (p.1).

US election security is shaped by three factors: (1) software vulnerability management constrained by outdated certification regimes; (2) inconsistent vendor transparency on vulnerabilities and patch status; (3) cybersecurity immaturity of many SLTT networks (p.1).

From 2019-2024 CISA partnered with Idaho National Laboratory (INL) on the Critical Product Evaluation program for direct technical assessments of election software, often before public release, upon vendor request (p.2).

Assessment methods included static source-code review, binary fuzzing of parsers/media handling/data import modules, cryptographic implementation analysis (PRNG misuse, poor key handling), interface/authentication/authorization/API testing, supply-chain dependency reviews, and dynamic analysis in adversarial runtime environments (p.2).

The Critical Product Evaluation program created disincentives for transparent, industry-standard vulnerability disclosure; it was retired in 2024 based on stakeholder feedback, with final reporting concluding in 2025 (p.3).

Vulnerabilities found included input-validation bugs, insecure deserialization, insufficient logging, race conditions, insecure crypto primitives, and privilege-escalation paths; CISA did not independently validate whether production builds in SLTT environments incorporated all fixes (p.3).

SLTT officials face update difficulties due to minimal IT budgets/staff, short deployment windows, and "lockdown" periods (sometimes mandated by state law) barring changes before election day; state laws may require only EAC (Election Assistance Commission)-certified software, delaying updates that postdate certification (p.3).

Technical risks: vendors cannot ship fixes outside narrow certification cycles without jeopardizing eligibility; third-party components (OS, drivers, middleware, crypto libraries) cannot be patched at modern cadence; legacy OS baselines accumulate unpatched vulnerabilities; absence of secure auto-update prevents rapid zero-day response (p.3).

CISA recommends national policymakers encourage harmonization of patch-management and certification rules (p.4).

Across penetration tests and red-team engagements, CISA observed flat/minimally segmented networks, weak identity and access management (poor MFA, shared credentials, weak service-account hygiene), lack of endpoint hardening, legacy remote-access/file-transfer pathways, and insufficient traffic monitoring (p.4).

In multiple cases CISA assessors gained full network control within hours or days, showing many SLTT partners remain "soft targets" (p.4).

Most states have moved away from paperless electronic voting machines; in 2020, ImageCast X Ballot Marking Devices printed ballots encoding selections in a barcode voters could not verify, and a researcher showed hackers could change the barcode-encoded votes without physical machine access (p.4).

Citation given: J. Alex Halderman, Security Analysis of Georgia's ImageCast X Ballot Marking Devices, expert report in Curling v. Raffensperger (Civil Action No. 1:17-CV-2989-AT, N.D. Ga.), with Drew Springall, July 1, 2021 (p.4).

ODNI commissioned a forensic examination of Dominion Voting Systems devices used in Puerto Rico's 2024 election; CISA reviewed the report but did not have access to the devices and could not perform its own examination (p.4).

Citation given: Dominion Voting Systems Democracy Suite Preliminary Vulnerability Assessment, Mojave Research for the ODNI, September 25, 2025 (p.4).

Vendor threat models assume strong segmentation between election systems (voter-registration databases, e-pollbooks, election management/tabulation systems, central scanning) and enterprise IT, but 2019-2024 assessments showed election systems reachable from enterprise hosts via shared auth domains, legacy VLANs, or "temporary" exceptions; poor firewall hygiene; co-location on general-purpose virtualization clusters; and unreliable "airgap" assumptions (vendor support tunnels, remote-management tools) (p.5).

CISA recommended mitigations: harmonize patch/certification rules; adhere to CISA Best Practices for Securing Election Systems; use human-readable paper ballots; conduct post-election manual audits of paper ballots before certification (p.5-6).

Additional recommendations: encourage vendors to assign CVE numbers, notify customers if source code is leaked/stolen, report incidents to authorities, include a software bill of materials (SBOM); and ensure transparent documentation of all security incidents and remediation (p.6).

Conclusion: US election-system security issues stem from complex interdependencies between vendors, certifiers, policymakers, and resource-constrained SLTT partners, not any single entity (p.6).