EPIC Database Used For Facial Recognition
When the scores in the match-up crossed the pre-set cut-offs, Aadhaar numbers were seeded into the rolls. The Supreme Court halted NERP-AP on August 11, 2015, and the Election Commission ordered state governments to suspend Aadhaar collection and seeding.
Hyderabad: Photographs given only for voter ID cards were all reused by the Telangana government to build facial-recognition tools, independent AI researcher and cybersecurity expert Srinivas Kodali has alleged in petitions to the State Election Commission (SEC) and the Chief Electoral Officer (CEO). In a 75-page complaint, he said the state government, in 2015, fed EPIC voter data into state systems without consent to develop its Real Time Digital Authentication of Identity (RTDAI), in violation of the Representation of the People Act, 1950.
Kodali alleged that under the National Electoral Roll Purification and Authentication Programme (NERP-AP) in 2015, the CEO’s office pushed EPIC data to the State Resident Data Hub (SRDH), matching names, parents’ names, dates of birth, addresses and photographs to Aadhaar using ‘Soundex’ and ‘Levenshtein distance’.
When the scores in the match-up crossed the pre-set cut-offs, Aadhaar numbers were seeded into the rolls. The Supreme Court halted NERP-AP on August 11, 2015, and the Election Commission ordered state governments to suspend Aadhaar collection and seeding.
Even so, booth level officers (BLOs) collected missing Aadhaar numbers in the field and “force seeding” followed, Kodali alleged. Deletion lists marked people as “duplicate”, “dead” or “shifted”. Many reached polling booths in 2018 to find their names missing.
By 2019, the same image base powered the RTDAI, he alleged, basing his claim on the data he sourced from Dr. MCR HRD Institute of Telangana.
EPIC voter photos were then used for facial-recognition checks on T-App Folio, Srinivas showed with what he called image proofs in his complaint.
The government describes RTDAI as “presence-less, contact-less, paper-less” and “three-factor verification” (liveness, face match, demographic check). Pensioners filed selfie life certificates. The transport department’s friendly electronic services in the FEST app used face match for licence renewals and issuing duplicate copies. The DOST system for admissions into degree colleges used the same check, the researcher said.
“There is no consent screen, no opt-out, and no independent check. Photos given only for voting now ‘gate’ pensions, licences, admissions, and, on pilot, voting itself,” Kodali told Deccan Chronicle.
He asked election authorities to stop sharing electoral data, delete data already shared, commission an independent inquiry, and issue a letter within 15 days explaining the legal authority for any sharing.
Asked about the claims, Telangana Chief Electoral Officer C. Sudarshan Reddy said: “We have not shared any such data. All the voter data is confined to the election officers and solely used by them. It has not been given to anyone.”
Important points:
2015: EC shares EPIC photos with State Resident Data Hub under National Electoral Roll Purification and Authentication (NERP-AP).
2015: Supreme Court halts NERP-AP; state keeps seeding Aadhaar.
2016-018: Aadhaar matches and dedupe runs; voter deletion lists issued.
December 2018: Voters find names missing at Assembly polling booths.
2019: Real Time Digital Authentication of Identity (RTDAI) app built; pension selfie life certificates on T-App Folio.
2020: RTA friendly electronic services (Fest app) and DOST degree college admissions add face match.
Jan. 2020: Voter ID pilot with facial recognition in municipal polls (Kompally).
2021+: E-voting pitch with blockchain; “Year of AI” push.
What happens when Electoral data is shared?:
Privacy invasion: Voter photos repurposed without consent.
Wrongful deletions: genuine voters removed pre-2018 polls.
Impersonation risk: Leaked photos can be used for fake voting.
Mass surveillance: one database links welfare, transport, elections.
Loss of trust: citizens unsure if vote is secret or secure.