"Technology-driven verification in welfare delivery can curb leakages but risks excluding genuine beneficiaries." Examine this with reference to face authentication under VB-G RAM G.

Under VB-G RAM G, the scheme that replaces MGNREGA, workers' attendance is recorded on the National Mobile Monitoring System (NMMS). Face authentication was added to it in March 2026, and an upgraded version built with IIT Madras was launched in October 2026 [1]. The app reduces fraud, but it also creates a new point where a genuine worker can fail the check.

How face authentication curbs leakages

• Ghost workers: Face authentication was brought in after complaints of fake attendance and of wages claimed in other people's names. Liveness detection rejects dummy images and old photographs [1].
• Layered checks: When AI de-duplication flags a photo, the wage list and Fund Transfer Order (FTO) are withheld automatically. Wages are then paid based on the e-MB, the engineers' measurement of work actually done, and sent by DBT [1].
• Scale: About 40 crore of roughly 46 crore person-days are face-authenticated, and nine states are at 100% [1].
• Device integrity: Play Integrity and anti-cloning checks block tampered apps, and states must prosecute misuse [1].

How it risks excluding genuine workers

• Payment is stopped first and checked later: A wrongly flagged worker waits for wages until an official clears the case. Even a 1% error rate on 40 crore person-days would hold up about 40 lakh days of wages (illustrative).
• Past evidence: In 2024 the Standing Committee found that workers who could not mark NMMS attendance went unpaid. It noted that many lacked smartphones or connectivity, and that 14% of job card holders were not Aadhaar-enabled [2].
• Delay compensation is rarely paid: In 2022-23, only ₹59 lakh of the ₹94 lakh approved was actually paid [2]. The cost of delay falls on the worker.
• Failures are unequal: Elderly and disabled workers depend on the exception route. The mate files it, and no deadline or appeal is stated [1].
• Blind spot: The app checks who came, not how much work was done. Inflated measurements can still leak money.

Way forward

• Publish state-wise data every month on photos wrongly flagged and how long clearing takes.
• Clear flagged FTOs within a fixed time and pay automatic delay compensation.
• Give attendance a routine backup, such as Gram Panchayat verification, the way ABPS payments can fall back on NACH [3].
• Let workers file grievances themselves, and strengthen social audits of the works.

Face authentication tackles a real leak, but it currently puts the cost of machine error on the poorest worker. Fallbacks, published error data and enforceable timelines would keep it proportionate, as Puttaswamy requires. They would also honour the right to work under Article 41: verification should protect wages, not deny them.