Coding agent reconstructed Georgia ballots and exposed voter choices
Review RAT-119 · Catalogue RAT-119
Princeton’s original states the human researcher deliberately gave a coding agent a published vulnerability and public records to reconstruct ballots. This is explicitly intended human-directed research, despite serious privacy implications. It is distinct from CAT-U08’s unrelated Georgia Data Hub aggregate-statistics access. Do not conflate either.
Reviewed: · Rule: roguewatch-organizational-harm-v1
Dates and discovery
- Incident date
- Not disclosed
- Added to research / record
- Posted to RogueWatch
- No linked public incident
- First observed
- 2026-10-30T13:13:38.595729Z UTC
- Discovery channel
- User-supplied comparison audit
- Discovery source
- https://rogueaitracker.com/incidents/coding-agent-georgia-ballot-reconstruction/
Dates use America/Los_Angeles. For older entries, Posted to RogueWatch uses the saved record addition date; a separate deployment time was not recorded. Future posting timestamps record first public qualification. Source publication and incident occurrence are separate.
Sources
- Rogue AI Tracker discovery entryReporting / interview · rogueaitracker.com
- blog.citp.princeton.eduOriginal account · blog.citp.princeton.edu
Decision history
- · Case-by-case comparison audit
RAT-119: out-of-scope. Princeton’s original states the human researcher deliberately gave a coding agent a published vulnerability and public records to reconstruct ballots. This is explicitly intended human-directed research, despite serious privacy implications. It is distinct from CAT-U08’s unrelated Georgia Data Hub aggregate-statistics access. Do not conflate either. Source checks: read: https://blog.citp.princeton.edu/2026/08/03/an-algorithmic-failure-beneath-the-secret-ballot/. Princeton’s original states the human researcher deliberately gave a coding agent a published vulnerability and public records to reconstruct ballots. This is explicitly intended human-directed research, despite serious privacy implications. It is distinct from CAT-U08’s unrelated Georgia Data Hub aggregate-statistics access. Do not conflate either.
Evidence source
A review decision applies only to the named scope. Confidence in a report and qualification under the inclusion rules are separate questions. Review records do not add incidents to the globe or leaderboard.