Georgia's latest ballot-image audit of the June 16 primary runoff turned up a grand total of 23 discrepancies out of more than 1.1 million ballot images, state officials announced Friday, and none of them were enough to budge a single race. The review backed up the certified winners in every runoff contest and found virtually no mismatch between what the scanners read and what the machines tallied. Every identified error showed up on hand-marked paper ballots, while ballots produced by ballot-marking devices tracked their records exactly, according to officials.
According to CBS News Atlanta, auditors examined all 1,111,856 ballot images from the runoff and detected just 23 discrepancies. In a release, the Georgia Secretary of State’s office put the audit's accuracy at 99.9979 percent and said every one of those discrepancies came from a hand-marked ballot, while the 1,079,408 ballots produced by ballot-marking devices showed zero mismatches. The office also noted that the ballot-image audit was carried out separately from a hand-count risk-limiting audit that likewise confirmed the certified results.
CBS News Atlanta reported that state officials projected the same hand-marked ballot discrepancy rate onto a high-turnout presidential election with about 5 million ballots, which would translate into roughly 3,500 discrepancies statewide. The outlet also quoted Secretary of State Brad Raffensperger saying his team "will continue to conduct audits after every election" as part of an ongoing effort to reassure voters about Georgia's vote-counting systems.
What the numbers show
The latest runoff review tracks closely with a broader ballot-image audit the state released in June. That earlier report, from the Georgia Secretary of State’s office, examined 2,081,900 ballots and found 159 discrepancies, again with most issues tied to how humans filled out or interpreted hand-marked ballots. State officials have pointed to those results to argue that using ballot-marking devices, then layering multiple post-election checks on top, significantly cuts the odds of tallying errors slipping through.
How the audit works and what comes next
Ballot-image audits rely on text-recognition software that reads the human-visible text on the scanned ballot images and compares that text to the tabulator tallies. That approach lets the state run a broad, data-driven check of results without launching a full manual recount, and it sits alongside risk-limiting audits and traditional hand counts. As described by GovTech, the tool is built to flag rare mismatches and help election officials spot places where training or procedures might need a tune-up.
For now, state officials say both the ballot-image audit and the risk-limiting audit point to the same conclusion: the June 16 runoff results were counted accurately. And if Raffensperger sticks to his word, audits like these are going to be a regular feature of Georgia elections for the foreseeable future.
