A Government Algorithm Declared Living People Dead and Cut Off Their Pensions

What happened

Starting around 2020, an AI-powered system used by the Indian state of Haryana began declaring welfare beneficiaries dead. The people it killed on paper were alive. Their pension payments stopped anyway.

The system is called Parivar Pehchan Patra, or PPP. It was built by the Haryana state government to assign families a unique eight-digit identifier based on income, age, employment, and related data, and to link that record to birth, death, and marriage registries. The stated goal was straightforward: streamline welfare delivery and cut fraud by maintaining a single source of truth across government databases. When a death was recorded anywhere in the system, PPP would automatically update the family record and halt related benefits. The problem is that it also did this when no death had actually occurred.

The scale of the error emerged publicly in January 2024, when government data showed that over 300,000 pensioners had had their benefits withheld. One case that drew particular attention involved a 102-year-old man named Dhuli Chand, who was forced to organize a mock wedding procession to demonstrate to local officials that he was, in fact, still breathing. The absurdity of the proof required of him was not an outlier; it was the process that the system's errors imposed on thousands of people with no other way to contest what a database had decided about them.

The failure had more than one source. PPP's linkage to death records meant that a mis-entry anywhere upstream, a keying error, a mismatched name, a record pulled from the wrong family, could cascade directly into a cut-off with no human review step intervening. The system also made predictions about income and employment as part of its eligibility logic, and reporting at the time indicated those predictions were frequently wrong. A model that was wrong about whether you had income, combined with one that thought you were dead, produced a result that the people affected had almost no institutional channel to correct.

This is the gap that systems like PPP expose most clearly. The algorithm issued a determination, the benefits stopped, and the person on the receiving end of that decision had no access to a provable record of what the system did, which input triggered the outcome, or who, if anyone, had reviewed it before it went into effect. Accountability infrastructure at that layer, not just an appeals window but a logged, auditable record of each decision and its basis, would have made each error visible and reversible at the moment it happened, rather than after a press cycle forced the government to release its own numbers.

Reported impact

Affected parties

Not publicly disclosed

Harm type

Not publicly disclosed

Scale

Not publicly disclosed

Financial impact

Not publicly disclosed

Regulatory action

Not publicly disclosed

Classification

Organization

Not publicly disclosed

AI system

Not publicly disclosed

Industry

Not publicly disclosed

Country

Not publicly disclosed

Provider

Not publicly disclosed

Incident type

Not publicly disclosed

Relevant governance controls

Governance control mapping is not available for this record.

• No controls mappedNot publicly disclosed

Control mapping is analytical. It does not state that any control would have prevented the incident.

Sources and evidence

This record was researched and written by the Index. The event is also catalogued in the following database, which is listed for cross-reference.