₹6.97 crore, one dead-patient roster, and the case for an audit trail that runs in real time.

The Comptroller and Auditor General's audit of Ayushman Bharat–PM-JAY surfaced findings uncomfortable for a scheme built on the promise of cashless, verified care. Nearly 7.5 lakh beneficiaries were found registered against a single placeholder mobile number — 9999999999 — with a further 1.39 lakh registered against 8888888888, and 96,000 against 9000000000. These aren't obscure edge cases; they're the kind of pattern a basic validation rule would catch on day one, at the point of registration, before a single claim is ever filed against them.

More striking: ₹6.97 crore was paid out for the treatment of 3,446 patients the scheme's own database had already recorded as deceased — 3,903 claims processed for people the system itself said were no longer alive. The audit also found no mechanism preventing the same patient from being "admitted" to two hospitals during the same period, and documented beneficiaries paying out of pocket for care the scheme was explicitly designed to make free. States including Chhattisgarh, Haryana, Jharkhand, Kerala, and Madhya Pradesh reported the highest volumes of these irregularities.

What none of this actually says about the scheme

None of this points to a scheme short on funding, intent, or scale — PM-JAY is one of the largest health assurance programmes in the world by beneficiaries covered, and the underlying ambition behind it is sound. It points to a gap between the moment a claim is approved and the moment anyone can independently verify it happened as described. That gap isn't a personnel failure or a corruption story in the way it's sometimes framed; a claims volume at PM-JAY's scale cannot be manually cross-checked against a deceased-patient roster claim by claim. It's a systems-design gap — the checks that would have caught a placeholder mobile number or a dead-patient claim simply weren't built into the pipeline the claim travels through.

Why a yearly audit structurally can't fix this

A yearly CAG audit can only ever find these patterns after the money has already moved and the claim has already been paid. That's not a criticism of the CAG's process — it's the nature of a periodic, sample-based review of a claims volume that size. The audit is doing exactly what an audit is designed to do: look backward, at scale, on a fixed cycle. The problem is that a scheme moving public money in real time needs a check that also runs in real time, not once a year, sitting alongside the audit rather than replacing it.

A programme built with a live accountability layer — validation, deduplication, and traceability designed in at Stage 01 of implementation, not appended at Stage 4 as a reporting feature once everything else is already built — finds these patterns before disbursement, not after. A duplicate placeholder mobile number gets flagged at registration. A claim against a beneficiary the system already marked deceased gets held automatically, before payment, for a human to actually look at. None of this requires new legal authority or a bigger audit team; it requires the same disbursement pipeline that pays a valid claim to also be the pipeline that catches an invalid one, on the way through rather than a year later.

This is the specific problem Financial Technology is built to solve, and it's why the Mandate-to-Ground Framework treats Accountability as a design decision made at Stage 01 alongside the scheme's core rules, rather than a dashboard bolted on once a programme is already live and its failure patterns are already baked in.