FULTON COUNTY — The FBI explored using artificial intelligence to assess the validity of signatures on tens of thousands of mail-in ballot envelopes seized from Fulton County, Georgia. The initiative is part of a broader effort tied to the Trump administration's investigation into the 2020 election, and involves comparing signatures on ballot envelopes with those on other election documents, such as voter registration forms.

In January, the FBI raided Fulton County and collected about 700 boxes of election materials. Among the seized items were approximately 150,000 mail-in ballots, roughly 116,000 of which were cast for Joe Biden. Biden defeated Donald Trump in Georgia in the 2020 election by 11,779 votes. The signature-matching initiative was under discussion as recently as late June, according to internal sources.

The FBI mandated that 260 analysts be redirected from field offices nationwide to focus on the Fulton County probe. Some FBI analysts were fired rather than work on the effort, highlighting internal tensions over the investigation’s direction. Analysts are also comparing a spreadsheet of 175,000 voters with a commercial database to determine if the voters are still alive, a step aimed at identifying potential irregularities.

An FBI technology specialist involved in the effort acknowledged the challenges of using artificial intelligence for signature verification. "It’s up to the builder of the system to define the guidelines," the specialist said. Another FBI staffer familiar with the probe added, "Everyone is of the opinion that, whether they find anything or not, they are going to continue to pursue proof of fraud."

Critics have raised concerns about the reliability of AI-driven signature analysis. Linton Mohammed, former president of the American Society of Questioned Document Examiners, questioned the feasibility of the approach. "Signatures are one of the most difficult forensic sciences, and I don’t think AI is going to be able to do this," he said. He added, "Signatures vary — unlike DNA or fingerprints."

Background information underscores the scientific and procedural challenges of signature verification. A 2009 National Academy of Sciences report concluded that "the scientific basis for handwriting comparisons needs to be strengthened." Additionally, signatures can change over time as people age or experience health events such as a stroke. These variations complicate forensic comparisons.

The limitations of signature matching have been documented in legal and academic settings. In a 2020 lawsuit challenging an Ohio signature-matching law, a political scientist testified that his analysis suggested 32 legitimate ballots were blocked for every illegitimate one. Georgia itself moved away from signature matching after the 2020 election: the state’s Election Integrity Act of 2021 replaced the practice with stricter verification methods, such as requiring mail-in voters to provide their driver’s license number or copies of licenses.

The FBI’s use of artificial intelligence to examine ballot envelope signatures represents a expansion of forensic techniques in election investigations. With 150,000 seized ballots—most favoring Biden—and an election decided in Georgia by fewer than 12,000 votes, even small-scale validation efforts could carry political weight. However, the scientific community has long questioned the reliability of handwriting and signature comparisons, and Georgia has already abandoned the method in favor of more standardized verification processes.

The probe reflects ongoing efforts to revisit the 2020 election results, even as internal FBI resistance—evidenced by staff resignations and reassignments—reveals institutional friction. The outcome of this analysis, whether it uncovers irregularities or not, may influence public trust in both election integrity and the evolving role of AI in law enforcement.