AI tools revive ballot-order recovery flaw in US voting scanners

A researcher exploited a four-year-old vulnerability in voting scanners used by 21 US states, recovering ballot-cast order in Georgia's May 2026 primary using AI tools and public data. The attack required no network access or source code, only public early-voting lists, raising concerns about voter privacy and election integrity. The finding is unverified independently but based on a credible security researcher's report.