rollcallAI, a Voter Registration Analytics Dashboard Launched

Latanya Sweeney, together with researchers and developers from the newly formed nonprofit the Tech Trust Engine and the Public Interest Tech Lab at Harvard University, developed an AI processing pipeline to systematically monitor real-time changes in voter registrations and report statistically significant shifts at the state, city, and ZIP code levels, as well as across various demographic groups. (rollcallai.org)

Measuring Erroneous Voter Registration Deletions

Latanya Sweeney and Josh Visnaw in the Public Interest Tech Lab at Harvard University conducted a study that found:

• In June 2024, Ohio officials publicly announced the scheduled removal of 158,857 voter registrations as part of its routine list maintenance process.
• At least 1 percent of those voters were erroneously removed and had to take corrective action to remain registered—introducing a statewide Erroneous Removal Rate (ERR) of 1 percent.
• Geographic disparities exist, with municipalities like Warren and Girard showing error rates of 7 to 9 percent—far above the statewide average.
• These findings underscore the critical need for continuous, data-informed monitoring of voter rolls to understand impacts on voters.
• techscience.org/a/2025041702

An Accounting of 100K Voter Registrations Dropped During the Quiet Period

Latanya Sweeney in the Public Interest Tech Lab at Harvard University conducted a study that found:

• Ohio election officials removed 109,119 voter registrations during the federally restricted 90-day "quiet period" leading up to the 2024 general election.
• While this study found 87% of the removals justified, no legally permissible explanation was found for 13%—or 14,539 registrations.
• The findings raised legal concerns and underscore the need for greater transparency and ongoing data monitoring to safeguard elections.
• techscience.org/a/2025041703

VoteFlare

VoteFlare (https://voteflare.org), an automated service that monitors voters’ registration records in real time and alerts them when any change or issue requires their attention, served as an early iteration of what eventually became rollcallAI. First deployed at scale during the 2020 Georgia Senate runoffs, VoteFlare monitored voters’ registrations and notified them of problems that could interfere with their ability to vote, providing an early proof of concept for voter-level monitoring. Building on that success, the system expanded to a 50-state model, demonstrating that automated registration checks could operate reliably across diverse election environments. Today, VoteFlare is live, and its underlying infrastructure and insights helped shape the nationwide analytics capabilities that rollcallAI now delivers.

Vulnerabilities found in Voter Websites

Latanya Sweeney, Jinyan Zang and Ji Su Yoo, in the Data Privacy Lab at Harvard University, led a study that found:

• Websites for 35 states and DC in 2016 were vulnerable to voter identity theft attacks: an imposter could submit changes to voter registration information
• An imposter needed a combination of voter's name, date of birth, gender, address, Social Security Number, or Driver's License Number
• Relevant data can be acquired from government, data brokers, or darknet markets. Total cost of an automated attack against 1% of all vulnerable voter registrations nationwide ranged from $10,081 to $24,926 depending on the data source used. States cost less, e.g., $1 for Alaska and $1,020 for Illinois
• A voter identity theft attack could alter up to 10 percent of election results by allowing imposters to submit address changes, delete voter registrations, or request absentee ballots online across a state—without detection.
• techscience.org/a/2017090601