Yahoo AI Exploit De-anonymizes 1.52 Million Georgia Ballots
Article Content
- •AI agent exploited a known vulnerability to de-anonymize 1.52 million ballots.
- •Georgia's election officials implemented emergency measures but face political stalemate.
- •The exploit highlights the risks of unpatched vulnerabilities in election systems.
An AI coding agent costing $20 successfully de-anonymized 1.52 million ballots from Georgia's May 2026 primary. The exploit utilized a known, unpatched vulnerability in Dominion voting machines, allowing the AI to match 98.9% of in-person ballots across 114 counties. This incident has prompted Georgia election officials to implement emergency measures, including redacting ballot ID numbers and shuffling ballots before printing. Despite these actions, a permanent technical solution remains stalled due to political disagreements in the state legislature. The vulnerability, disclosed in 2022, involves deterministic identifiers assigned to electronic ballot records, with a software fix certified in March 2023 but not widely deployed. This event marks a significant shift in election data transparency policy, raising concerns about ballot privacy ahead of the upcoming midterm elections.
Ask AI about this cluster
Answers cite the sources they use
Timeline
More articles in this cluster (4)
Following this threat?
Track Election Assistance Commission in your own feed — you're alerted when they show up in new reporting, leak sites or exploitation.
Free account · no card needed
Common questions
What vulnerability was exploited?
What measures has Georgia taken in response?
Is there a permanent fix for this issue?
Continue Reading
Critical Authentication Bypass in Cisco Catalyst SD-WAN Manager Exploited On September 30, 2026, Cisco disclosed a critical vulnerability (CVE-2026-76504) in the Catalyst SD-WAN Manager that allows unauthenticated remote attackers to bypass authentication and gain admin-level access to the system. This flaw stems from improper handling of URI encoding in HTTP requests, enabling attackers to…