The FBI has weighed using artificial intelligence to cross-reference tens of thousands of mail-in ballot signatures seized from Fulton County, Georgia. Internal communications reviewed by ProPublica indicate that discussions regarding this forensic initiative took place as recently as late June, targeting a direct comparison between ballot envelope signatures and official registration records.
A Massive Federal Audit of Fulton County Ballots
The bureau’s involvement stems from a January raid on the Democratic stronghold, which resulted in the seizure of approximately 700 boxes of election materials, including roughly 150,000 mail-in ballots. To process this volume, the FBI reassigned 260 analysts from field offices across the country.
The investigation extends beyond handwriting analysis. According to an agency memo, the bureau is currently vetting a spreadsheet of 175,000 voters against commercial databases to determine their status, specifically flagging whether any individuals were deceased when they cast their ballots.
The Forensic Limitations of Automated Handwriting Analysis
Deploying AI for forensic handwriting verification remains a deeply debated strategy. A technology specialist familiar with the FBI’s work stated that while high-quality datasets could allow for “somewhat accurate” analysis, the system’s output is entirely dependent on the guidelines set by its developers.
Experts warn that signatures lack the static reliability of DNA or fingerprints. Linton Mohammed, a former president of the American Society of Questioned Document Examiners, noted that signatures are “one of the most difficult forensic sciences, and I don’t think AI is going to be able to do this” because “signatures vary — unlike DNA or fingerprints.” David Gerber, a senior vice president at ParaScript, added that even with high-performing software, trained specialists should review approximately 10% of cases to ensure accuracy.
Legislative Fallout from 2020 Election Claims
This federal inquiry arrives after Donald Trump’s repeated claims that signature verification failures cost him the state of Georgia. Those allegations prompted the passage of the Election Integrity Act of 2021, which moved the state away from subjective signature matching toward stricter ID requirements, such as driver’s license numbers.
Critics of signature-matching processes point to a history of inequity. Research shows these methods often disproportionately reject ballots from the elderly, voters of color, and those with disabilities. A 2009 report by the National Academy of Sciences underscored this, highlighting that the scientific basis for handwriting comparison requires further strengthening.
Internal Bureau Friction and Uncertain Outcomes
The future of the FBI’s signature-matching project is unclear. Within the bureau, some staffers have attempted to address political pressure by highlighting the technical limitations of broad signature analysis, yet agency leadership has continued to push the probe forward.
Deep-seated concerns persist regarding the potential for political influence on the investigation. Neither the White House nor the FBI provided comment on the project’s current standing.
Keep reading