1 hr ago
Telangana SIR Anomalies Breakdown Unshared Amid AI Accuracy Concerns
Election officials found possible data problems affecting nearly 93 lakh voters in Telangana.
These problems were identified during a special voter-list review called SIR.
The state election office has not received a detailed list showing how many problems belong to each category.
An official said most issues may simply be spelling mistakes.
Other issues involve unusual ages or incorrect family connections.
Some experts are worried that computer systems using AI may make mistakes, especially across different languages.
They say people should check every computer-generated warning.
They also want voters to have a fair chance to correct the information.
Nearly 93 lakh Telangana voters were flagged for logical discrepancies during the SIR exercise.
The Election Commission of India has not provided the CEO’s office a category-wise anomaly breakdown.
An official estimated that 70–80% of anomalies involve spelling mistakes, with others involving age and family-linking discrepancies.
Reported anomalies include implausible age or sibling gaps, inconsistent parent details, missing documents and Aadhaar-only records.
Researchers and activists questioned AI reliability and urged human verification, transparency and due process before voters face consequences.
- Who
- The Election Commission of India, Telangana’s Chief Electoral Officer’s office, technology researcher Srinivas Kodali and SIR activist M.A. Mujeeb Ayyub.
- What
- Nearly 93 lakh voters were flagged for logical discrepancies, but the category-wise breakdown has not been shared with Telangana’s CEO office.
- Where
- Telangana, with the report datelined Hyderabad.
- When
- During the special intensive revision referred to as SIR 2026; the article does not provide specific dates.
- Why
- The discrepancies were identified during electoral-roll checks involving spelling, ages, family relationships, documents and other voter records.
Election Officials’ Account
Researchers’ and Activists’ Concerns
Nature of the anomalies
Election Officials’ Account
A higher official said roughly 70–80% of flagged cases are spelling mistakes, while the rest involve age gaps and family-linking discrepancies.
Researchers’ and Activists’ Concerns
Critics argued that algorithmic errors, particularly across different languages, could produce inaccurate flags.
Role of AI
Election Officials’ Account
The article describes ERONET 2.0 as the centralized platform used for electoral-roll management and reports questions about AI being used in the SIR process.
Researchers’ and Activists’ Concerns
Srinivas Kodali and M.A. Mujeeb Ayyub questioned the reliability of probabilistic AI systems and called for disclosure of their methodology and safeguards.
Voter safeguards
Election Officials’ Account
The CEO’s office has not received a category-wise breakdown from the ECI, and the official said the office had not sought it.
Researchers’ and Activists’ Concerns
Ayyub said every AI-generated flag should undergo human verification, transparency, due process and an effective opportunity for voters to respond.
Key facts
- Voters flagged
- Nearly 93 lakh Telangana voters
- Estimated spelling issues
- About 70–80% of anomalies, according to a higher official
- Other discrepancies
- Age gaps, progeny mapping and inconsistencies between current and previous rolls
- Breakdown status
- The ECI has not shared a category-wise breakdown with the CEO’s office
- Election platform
- ERONET 2.0 is the ECI’s centralized web and mobile-based electoral-roll management platform
- Technology concern
- Critics questioned the reliability and transparency of AI-assisted discrepancy detection
- Development partner
- Tata Consultancy Services collaborated with the ECI to develop ERONET 2.0, according to its 2024–25 annual report
Quotes
A higher Telangana election official
An unnamed senior official in Telangana's election administration
“AI for SIR runs on a ‘probabilistic system’. If you pose a question, the answer you get will be different, which completely depends on the percentage of AI’s predictions. The accuracy of the algorithm of ECI has always been challenged. The same model has been adopted for different languages creating more inaccuracies.”
deccanchronicle.com
“The ECI has not shared the break-up of the anomalies with us, nor have we sought them. Close to 70 to 80 per cent are spelling mistakes, while the remaining are age gap and logical discrepancies in progeny mapping.”
deccanchronicle.com
M.A. Mujeeb Ayyub
Advocate and SIR activist
“An algorithmic error must never become a reason for disenfranchisement. In the world’s largest democracy, every AI-generated flag must be subject to human verification, transparency, due process and an effective opportunity to respond. ECI must disclose the scale, methodology and safeguards governing AI use in SIR 2026. Technology should assist election officials — not decide who gets to exercise the right to vote.”
deccanchronicle.com
Sources
Related news
Telangana Poll Chief Among 93 Lakh Voters Served Notices
Telangana Delivers One-Third of Electoral Roll Anomaly Notices
Delhi SIR May Exclude 47.7 Lakh Voters From Draft Roll
Supreme Court Seeks Clarity on Delhi SIR Notices
Karnataka SIR Flags Nilekani, Kamaths and Other Voters
