AI exposes vulnerability in Georgia primary voting system, prompting urgent committee review

Researcher Max Springer from Princeton reported that artificial intelligence can combine publicly accessible voting proofs with ballot numbers to reconstruct the order of ballot submission in Georgia's recent primaries. By obtaining digital verification records from selected precincts, he showed that the supposedly random numbering can be decoded, allowing the reconstruction of voting sequences in 114 of 139 counties examined.

The analysis also suggested that, under low-traffic conditions such as early voting in small districts, the AI could potentially match voters to specific ballots using auxiliary logs. Georgia officials, including Secretary of State Brad Raffensperger, responded by ordering the redaction of identifying numbers on the records and scheduling an emergency committee session to address the security gap. Committee members debated whether manual shuffling of ballots before scanning could mitigate the risk.