PIB TrackerLiveness detection on a muster roll, with an exception for the old
The rural employment attendance app has been rebuilt with anti-tampering and AI face verification after complaints of fake attendance - and with offline recording and an exception route so that verification does not become exclusion.
What happened
- The Rural Development Minister launched an enhanced NMMS app for attendance under VB-G RAM G, built with IIT Madras support, adding device-level security, AI-based face verification, Google Play Integrity, anti-cloning and anti-tampering safeguards and liveness detection.
- It provides offline attendance recording where there is no network, uploaded after the muster roll closes, and an authorised exception mechanism where face authentication fails for elderly workers or persons with disabilities or a device malfunctions.
- The stated reason: complaints of fake attendance and wages claimed in other people's names. Face authentication was introduced in March 2026, and nine States - Andhra Pradesh, Karnataka, Kerala, Odisha, Tamil Nadu, Telangana, Chhattisgarh, Rajasthan and Uttarakhand - report 100 per cent attendance through it.
- Scale: implemented across all rural areas from 1 July 2026, with work at 31.23 lakh worksites across over 2.41 lakh Gram Panchayats, more than 3.65 crore workers from 2.89 crore households, and nearly 46 crore person-days, of which about 40 crore were face-authenticated.
- Finance: Rs 95,692 crore allocated by the Centre for 2026-27, total provision with States above Rs 1.51 lakh crore, and a mother sanction of Rs 26,126 crore released as the first instalment.
For Prelims
- NMMS - the National Mobile Monitoring System - requires the mate at a worksite to upload time-stamped, geo-tagged photographs of workers during the day; it was introduced under MGNREGS in 2021, extended progressively to all worksites, and carried over to VB-G RAM G. Attendance on the app is the precondition for the muster roll and therefore for the wage payment.
- The governing law changed this year. VB-G RAM G is the Viksit Bharat - Guarantee for Rozgar and Ajeevika Mission (Gramin) Act, 2025, in force across the country from 1 July 2026, which replaced MGNREGA and raised the statutory guarantee of wage employment from 100 to 125 days a financial year for rural households whose adult members volunteer for unskilled manual work; Section 37 carries the transitional provisions. The programme remains demand-driven.
- Payment architecture: wages are paid through the National Electronic Fund Management System (NeFMS) and, since 2023, exclusively through Aadhaar Based Payment Bridge (ABPS) for eligible workers - the change that generated the standing debate about exclusion through technical ineligibility.
- Liveness detection distinguishes a live face from a photograph or replayed video, usually by requiring micro-movements or by analysing texture and depth; it is the standard countermeasure to the "presentation attack" the Minister describes as fake attendance.
- Why face recognition fails unevenly: error rates rise with age, with manual labour that wears the skin, and with poor lighting - which is precisely the population a rural works programme employs, and why an authorised exception route is a design requirement rather than a concession.
- Why the two money figures differ: the Centre's Rs 95,692 crore is its own allocation while the Rs 1.51 lakh crore includes the States' share - the cost of a wage-employment guarantee has always been split, with the Union carrying the unskilled wage bill and States a share of material costs and any unemployment allowance.
- Social audit by independent Social Audit Units, a district Ombudsperson and Gram Sabha-level scrutiny are the statutory accountability mechanisms of a rural employment guarantee; technology supplements them and does not replace them - an app log is not an audit.
- Comparable biometric pivots: ration delivery under the National Food Security Act moved to ePoS and Aadhaar authentication with the same trade-off - leakage falls, and genuine beneficiaries who fail authentication are excluded unless an override exists.
For UPSC: A precise example of the accountability-versus-exclusion trade-off in welfare delivery, with the scheme's own numbers attached. Deploy it on e-governance and DBT, on the rural employment guarantee, and on the use of biometrics in entitlement programmes.
What it is NOT: No national face-authentication rate - nine States are named at 100 per cent, the other twenty-odd are not - and no count of how many workers have been unable to authenticate, how many exception requests have been filed or how many were approved, which is the only way to measure exclusion. No data on attendance rejected, wages delayed or the value of fraud detected since March 2026, so the gain side of the trade-off is also unquantified. Nothing on grievance redress for a worker whose attendance is not recorded, and no reference to the Act's own social audit machinery.
For Mains
Syllabus: GS2.15 · GS2.10 · Linkage L1
Anchor
The Rural Development Ministry has launched a rebuilt NMMS attendance app for VB-G RAM G with liveness detection, AI face verification and anti-tampering safeguards, alongside offline recording and an authorised exception mechanism for workers whom face authentication fails.
Substantiation (data)
The scheme has provided work at 31.23 lakh worksites across more than 2.41 lakh Gram Panchayats to over 3.65 crore workers from 2.89 crore households since 1 July 2026, generating nearly 46 crore person-days of which about 40 crore were recorded through face authentication. Nine States report 100 per cent face-authenticated attendance. The Centre has allocated Rs 95,692 crore for 2026-27 against a total provision above Rs 1.51 lakh crore, with a first instalment of Rs 26,126 crore released.
Position
The design is better than the usual version of this story. Fake attendance is real, and a camera at the worksite is a proportionate answer to it - but the ministry has also built the two things such systems normally lack: offline capture where there is no network, and an authorised override when the face does not match. Recognising that elderly workers and persons with disabilities fail authentication more often, and writing a route for them into the system rather than leaving it to a local official, is the difference between verification and exclusion.
Counterpoint
None of it is measured. The release gives no national authentication rate, no count of failed authentications, no number of exception requests filed or granted, and no figure for the fraud the system has actually caught - so both sides of the trade-off are asserted. An override that exists on paper and is used by nobody looks identical in the data to one that is not needed, and with 40 crore person-days already flowing through face authentication, the number of people who could not be matched is not a detail.
Way forward
Publish monthly the authentication failure rate by State and age group, the exception requests filed and approved, and the fraud value detected since March 2026. Keep the muster roll valid when the app fails, and audit a sample of exception cases through the Act's own social audit units rather than through the app's logs.
Conclusion
A technically careful upgrade to the gate between 3.65 crore workers and their wages. The exception mechanism is the right instinct; publishing how often it is used is what would prove it.
Deploys into: E-governance and biometric verification in welfare · Rural employment guarantee · DBT and exclusion errors · Accountability mechanisms in schemes
Ministry of Rural Development · 2026-10-23 · PRID 2327556 · PIB source ↗