← Voltar ao feed

Analysis: US voting machine certification gaps, air-gap bypass risks and AI-driven election influence

1 fonte(s) · 1 atualização(ões) · publicado 29/10/2026 12:00 · coletado há 3 dias

SecurityWeek examines the cybersecurity posture of the US election process, focusing on ballot-marking devices and ballot-scanning/counting systems. Experts argue that EAC certification is decoupled from the four-year election cycle and can leave machines running software unpatched for two or more years, which Shai Bargil (Sequent) calls "pretty much meaningless." The article notes that essentially every voting machine brought to the DEF CON Voting Village has been compromised with hands-on access, and that air gaps can be bridged by a physically present attacker with a USB stick. A further concern is that counting relies on barcodes/QR codes that voters cannot visually verify, even though a human-readable paper ballot exists. Experts from Rapid7, Bugcrowd, Arctic Wolf, Doppel and NRLabs stress that the real adversary focus is the perimeter — voter registration databases, county websites and election staff email — plus AI-enabled influence operations. No evidence exists that machine tampering has changed a US election outcome.

Detalhes técnicos

  • Certification lag: EAC certification can take up to ~2 years and is not synchronized with the election cycle, leaving fielded devices unpatched.
  • Air-gap bypass: physical access at a polling place (e.g., malicious USB) could persist for all subsequent voters at that device.
  • Verification gap: counting machines read barcode/QR data, not the human-readable text, so voters cannot verify what is actually tallied.
  • Influence operations: referenced attributions include Russian 2016 influence activity (CIA/NSA/FBI, Mueller, Senate Intelligence Committee) and an Iran-attributed 2024 hack-and-leak against the Trump campaign.
  • AI abuse: the 2024 New Hampshire deepfake Biden robocall urging Democrats to stay home; hypothetical scenarios include Wikipedia vandalism feeding search engines (the Aug 12, 2026 Google/Sam Altman incident) and amplified false rumors on X, Facebook and LinkedIn.

Remediação

  • Treat voting machines as one untrusted component in a hostile system; assume vulnerable firmware, supply-chain targeting and compromised admin accounts (Christiaan Beek, Rapid7).
  • Open-source voting machine software to allow independent review and pentesting while retaining EAC certification.
  • Add an encrypted, independently verifiable vote record alongside the paper ballot; homomorphic encryption as in Microsoft ElectionGuard.
  • Maintain layered safeguards: pre-election testing, chain-of-custody, bipartisan poll workers, risk-limiting post-election audits, verifiable backups and redundancy.
  • Defend the perimeter: voter registration databases, county websites, campaign and election worker email accounts.

Linha do tempo

  1. securityweek.com ↗ · pub. 29/10/2026 12:00 · coletado 29/10 13:00 primeiro registro

    Analysis: US voting machine certification gaps, air-gap bypass risks and AI-driven election influence

    ler a análise desta fonte

    SecurityWeek examines the cybersecurity posture of the US election process, focusing on ballot-marking devices and ballot-scanning/counting systems. Experts argue that EAC certification is decoupled from the four-year election cycle and can leave machines running software unpatched for two or more years, which Shai Bargil (Sequent) calls "pretty much meaningless." The article notes that essentially every voting machine brought to the DEF CON Voting Village has been compromised with hands-on access, and that air gaps can be bridged by a physically present attacker with a USB stick. A further concern is that counting relies on barcodes/QR codes that voters cannot visually verify, even though a human-readable paper ballot exists. Experts from Rapid7, Bugcrowd, Arctic Wolf, Doppel and NRLabs stress that the real adversary focus is the perimeter — voter registration databases, county websites and election staff email — plus AI-enabled influence operations. No evidence exists that machine tampering has changed a US election outcome.

    Technical Details

    • Certification lag: EAC certification can take up to ~2 years and is not synchronized with the election cycle, leaving fielded devices unpatched.
    • Air-gap bypass: physical access at a polling place (e.g., malicious USB) could persist for all subsequent voters at that device.
    • Verification gap: counting machines read barcode/QR data, not the human-readable text, so voters cannot verify what is actually tallied.
    • Influence operations: referenced attributions include Russian 2016 influence activity (CIA/NSA/FBI, Mueller, Senate Intelligence Committee) and an Iran-attributed 2024 hack-and-leak against the Trump campaign.
    • AI abuse: the 2024 New Hampshire deepfake Biden robocall urging Democrats to stay home; hypothetical scenarios include Wikipedia vandalism feeding search engines (the Aug 12, 2026 Google/Sam Altman incident) and amplified false rumors on X, Facebook and LinkedIn.

    Remediation

    • Treat voting machines as one untrusted component in a hostile system; assume vulnerable firmware, supply-chain targeting and compromised admin accounts (Christiaan Beek, Rapid7).
    • Open-source voting machine software to allow independent review and pentesting while retaining EAC certification.
    • Add an encrypted, independently verifiable vote record alongside the paper ballot; homomorphic encryption as in Microsoft ElectionGuard.
    • Maintain layered safeguards: pre-election testing, chain-of-custody, bipartisan poll workers, risk-limiting post-election audits, verifiable backups and redundancy.
    • Defend the perimeter: voter registration databases, county websites, campaign and election worker email accounts.