Georgia Elections Board Scrambles After AI Linked 1.52 Million Ballots to Voters
Updated
Updated · The Guardian · Oct 2
Georgia Elections Board Scrambles After AI Linked 1.52 Million Ballots to Voters
3 articles · Updated · The Guardian · Oct 2
Summary
Less than two weeks before early voting, Georgia’s elections board held an emergency meeting after a Princeton researcher showed a $20 AI tool could reconstruct voting order for 1.52 million ballots—98.9% of in-person ballots in 114 counties.
Using public sign-in records and cast-vote files, Max Springer said the system’s ballot identifiers were predictable enough to match many voters to specific ballots, with secrecy “entirely lost” in some small counties.
Brad Raffensperger’s office responded by ordering ballot ID numbers redacted from post-election tabulation data, a step cryptographer Ben Adida said substantially fixes the flaw.
That fix still faces legal and operational pushback: some board members argued redaction may conflict with Georgia’s transparency rules, while a proposal to shuffle ballots before scanning was criticized as disruptive and risky.
Georgia officials said they had sought funding for a broader software overhaul for three years, but with early voting imminent, they warned any larger change is now too late and roughly $30 million short.