“This time I am not even wearing lipstick, what is it now?”
Avni’s voice, laced with frustration, echoed through the cramped, humid room of an Anganwadi centre in Delhi. Eight months pregnant and returning for the third consecutive day, she stared into the lens of an aging Android phone, hoping the app would finally recognise her face and allow her to collect the nutrition packet she desperately needed.
The walls around her were covered in crayon drawings and posters of government schemes promising nutrition and care. But between Avni and that packet of panjiri, an energy-rich nutritional supplement for expecting mothers, stood a technological barrier that seemed insurmountable: facial recognition technology that couldn't tell she was the same woman whose photograph existed in the system.
Behind the phone was Meena (name changed), an Anganwadi worker for over 18 years, now navigating not just the distribution of food packets but malfunctioning apps and digital verification errors that has turned her job into a daily struggle.
“Keep your face straight. Blink once. Maybe pull your dupatta over your head like in the earlier photo,” Meena said gently, trying to coax the system into compliance.
This wasn’t random advice. Avni’s photo, already stored in the Poshan Tracker—the government app launched in 2021 to monitor nutrition and health service delivery—shows her with a blue dupatta covering her head. The app attempts to match her current appearance with that reference image in real time, using the Facial Recognition Technology (FRT).
Since July 1, the use of FRT has been made mandatory to verify identities before distributing Take Home Ration (THR) under the Integrated Child Development Services (ICDS) scheme. This distribution happens twice a month, at 15-day intervals.
Implemented through Anganwadi centres, the ICDS is a flagship welfare programme that offers services like supplementary nutrition, preschool education for children aged 3 to 6, regular health check-ups, immunisations, and nutrition and health education for women. Its primary beneficiaries are pregnant women, lactating mothers and children under the age of six.
For children aged 6 months to 3 years, as well as pregnant and lactating mothers—and in some cases, adolescent girls and severely underweight children—the scheme provides THR. These food packages usually include fortified items such as panjiri (a wheat-based mix with ghee and dry fruits) for pregnant women, blended flours (like wheat, gram, or soy) for lactating mothers, ready-to-eat mixes and energy-dense supplements rich in essential micronutrients for children. The aim is to close nutritional gaps in vulnerable households.
The facial data, however, doesn’t just concern the women. For children below the age of six, the parent’s face—usually the mother’s—is used for verification. A mismatch means the child may be denied access to food or services.
“There is a problem with your camera, not my face,” Avni snapped. “I’ve just gained some pregnancy weight. My eyes and nose haven’t changed!”
The Digital Maze Of Welfare
Eventually, after several failed attempts and adjustments to lighting and angle, the system finally verified Avni's identity. Meena sighed in relief
“Thank god she had her Aadhaar eKYC done. Otherwise we would be stuck waiting for an OTP on her mobile number and with this terrible internet, that could’ve taken forever,” Meena muttered, already preparing the next file.
The verification process relies on Aadhaar eKYC (Electronic Know Your Customer), a digital identity verification system that uses data linked to India's 12-digit biometric ID. For Anganwadi beneficiaries, this system connects to the ICDS database, which stores Aadhaar details for all registered individuals. When someone's Aadhaar isn't updated or their eKYC is incomplete, the system sends an OTP to the mobile number stored in ICDS records.
For such beneficiaries, verification is only considered successful when both facial recognition and OTP authentication are completed, a process that often becomes a major hurdle in areas with poor mobile connectivity.
Poshan Tracker App Interface
Bureaucracy Over Benefits
Since the Union Ministry of Women and Child Development issued this directive in late June, daily struggles like Avni’s have become a commonplace across Anganwadi centres nationwide.
“It’s not like we work in banks and are handing out lakhs of rupees,” said Shalu (name changed), another Anganwadi worker, her voice carrying both fatigue and frustration. “It’s just a packet of raw food.”
The contrast between the technology's complexity and the simplicity of what it's protecting—basic nutrition for India's most vulnerable—strikes many workers as absurd.
The burden of this "digital compliance" extends far beyond working hours. “We try and try till the app accepts the face. If it doesn’t, we ask the woman to come the next day. Earlier, all this would take half a day. Now, just feeding data and verifying identities stretches into the evening,” said Shalu.
Shalu recalled a recent case where she had to shift a beneficiary from the pregnant to the lactating category after childbirth. “Her face had changed with dark spots and tiredness. The app just wouldn’t recognise her,” she said with a tired laugh. “I went to her home after work three times. The FRT failed each time. In the end, I just gave her the food and made a manual entry, hoping my supervisor understands.”
Take Home Ration packets for beneficiaries
The stakes are vastly different for workers and beneficiaries, Shalu explained. "For the beneficiary, it's just a packet of food, and often, it doesn't feel worth the hassle of coming back again and again just to prove who they are." Many women simply give up after repeated authentication failures.
"But for us, it's not just a packet, it's a data point. If too many go unrecorded, we get questioned by our supervisors." In the long run, consistent shortfalls in numbers could even put their jobs at risk.
For many Anganwadi workers, it's not just the glitchiness of the system but also its rigidity. “Earlier, we had an option of verifying manually. Now there’s no room for flexibility. This compulsion is driving us berserk,” Shalu said, squinting at the screen of an app that has gradually become the bane of their daily routine.
“Some of us are working late into the evenings, seven days a week,” said Shalu. “Even people in high-paying corporate jobs don’t work this much.”
Like many others, Shalu is currently managing two Anganwadi centres on her own. This added burden stems from a severe staff crunch after over 800 Anganwadi workers were allegedly terminated arbitrarily for participating in protests in 2022. While the court has since ordered their reinstatement, around 145 workers are still waiting to be brought back, leaving the remaining staff stretched thin.
A Shifting Burden of the State
Even the task of obtaining consent—central to any data protection framework—is being offloaded onto Anganwadi workers.
Each time a beneficiary’s face is to be authenticated through the Poshan Tracker, a consent screen appears stating that personal details, Aadhaar data, and facial images will be collected, stored, and used by state and central governments. The Anganwadi worker must then confirm: “I agree that I have read the above terms and conditions to the beneficiary and obtained their consent.”
In practice, overwhelmed and untrained, workers often tap ‘Accept’ without explaining the terms. They treat it as routine bureaucracy rather than a significant privacy decision.
Disha Verma, a human rights and technology researcher, told Decode this model unfairly shifts the burden of informed consent to frontline workers. “The state is outsourcing its responsibility,” she said. “Anganwadi workers are acting as agents of the state without being made aware of what that role entails.”
Most workers are unaware that they’re part of a sensitive data collection process. “If there is a breach or scam, people won’t blame a government server but the worker who took their details.”
Verma noted that while schemes like ICDS are increasingly data-driven, the women keeping them running are handling complex data work with no training, recognition, or incentive.“The technology is expensive. Why invest in it when it neither helps the worker nor the beneficiary while more pressing gaps remain unaddressed?”
The system's problems run deeper than mere technical glitches. “Lighting affects accuracy, facial features change—especially during and after pregnancy—and FRT often fails with elderly or dark-skinned individuals,” she said.
But the bigger concern, the researcher warned, lies in what happens if the technology does become highly accurate.
“That opens the door to surveillance risks, especially when women’s photos are stored without oversight. There’s even the potential for misuse, like deepfakes.”
Ultimately, she said, the question is not whether FRT can be improved but whether it should be used for welfare at all. In such a power-imbalanced relationship, consent becomes meaningless.
“If a beneficiary refuses, the Anganwadi loses a data point, misses a target, and potentially loses incentives,” Verma said. “That’s not real consent. It's coerced compliance.”
The Hidden Cost of Digital Welfare
The transformation of India's welfare system reflects a broader shift from service delivery to data collection. There are over 14 lakh Anganwadi workers across India, the backbone of what was once known as ICDS—a welfare programme launched in 1975, and rebranded as Mission POSHAN 2.0 in 2018. It is one of the largest state-run early childhood care initiatives in the world, addressing nutrition, health, and education for millions of children and mothers.
For decades, Anganwadi workers maintained handwritten registers to record attendance, nutrition status and supplies. The shift to digital tools like the Poshan Tracker was pitched as a way to modernise the system. But instead of reducing workload, it has added layers of bureaucracy and technical burden, with real-time monitoring of workers themselves becoming a central feature.
Now, facial recognition has further entangled this already tech-heavy workflow. The app, designed to track progress, has become a source of stress, especially for women like Meena and Shalu, who have had little training and even fewer support systems when the system breaks down.
According to Akriti Chaudhary, a union member of the Delhi State Anganwadi Workers and Helpers Union, the rollout of facial recognition is not just about digitisation—it may also be laying the groundwork for something more structural: privatisation by stealth.
She pointed out that over the past few months, several Anganwadi centres in Delhi have already been merged with private NGOs and schools, citing a ‘decline in number of beneficiaries’ as the reason. But that drop, she argued, can also be a direct consequence of the app’s failure to authenticate faces, which leads to undercounting.
“When workers fail to register beneficiaries because the facial recognition doesn’t work, it doesn’t matter how many hours they put in on the ground,” Chaudhary said. Supervisors and administrators only see the incomplete data on the app—not the repeated attempts, the home visits or the connectivity issues.
This discrepancy, she warned, risks being misinterpreted as inefficiency.
“It becomes easy then to say: ‘Look, this Anganwadi has too few beneficiaries, let’s shut it or merge it.’ And eventually, this will be used to justify laying off Anganwadi workers altogether.”
In her view, the growing reliance on facial recognition and other digital compliance tools—without adequate support or fallback systems—is not merely a case of flawed implementation, but a deliberate erosion of public systems.
“The tech becomes a filter,” she said, “and anything or anyone that can’t pass through it is treated as obsolete.”
For workers like Meena and Shalu, the concern is both practical and pressing. Data privacy may be a larger debate, but on the ground, the reality is more immediate: if the face match fails, they lose a beneficiary. What is framed as a technological upgrade often feels more like a daily gamble than real progress.
The irony is stark: a system designed to ensure welfare reaches its intended recipients is instead creating new barriers to access. Meanwhile, the workers tasked with implementing this system—predominantly women from marginalised communities themselves—bear the brunt of its failures. They work longer hours, face increased scrutiny, and carry the burden of a system that seems designed to fail both them and the people they serve.
As another mother entered the centre, balancing a sleepy toddler on her hip, Meena reopened the app. A few taps, a blurry photo, and the system beeped red. “Let’s try again,” she said. Another face, another glitch, another day in the life of digital welfare.
