EssenceالعربيةالعربيةRequest information
Insights

Over State Pension age, not a date of birth: proportionate sharing in practice

GovernmentTechnical

A service that needs to know someone is over State Pension age rarely needs their date of birth. Essence passes the answer, not the record: only the fields asked for, or only a signed answer.

A service needs to know that someone has reached State Pension age. A support scheme needs to know that a household qualifies. The usual way is to ask for the record behind the answer: a date of birth, a benefit letter, an account of someone's circumstances. The service then holds far more than its question needed, and for people in vulnerable circumstances that extra carries real risk.

What the vulnerabilities data standards set out

In June 2026 the Government Digital Service, the Department for Science, Innovation and Technology and the Department for Digital, Culture, Media and Sport published an alpha release of "standards for data sharing around vulnerable people and households". They were produced by a cross-government Vulnerabilities Working Group, which the Data Standards Authority established, building on models from the SAVVI programme. GOV.UK says they are intended for "information systems for the identification and support of vulnerable people or households, in particular for information exchange between different systems or organisations."

The release includes data models and a taxonomy of risk factors. One stated purpose of the taxonomy is "allowing the minimal data required to be shared on vulnerable individuals or households (for example, sharing “Person is Under 18” rather than date of birth) to minimise data sharing risk." The data models make the same point about privacy and security: "only sufficient information is shared at each stage".

The alpha feedback period closed in September 2026. The publishers say the material "will remain available for reference while we review the feedback received and consider it in the next phase of development."

The answer, not the record

The standard's example carries over to older people. GOV.UK describes State Pension age as "the earliest age you can start receiving your State Pension", and notes that it "is regularly reviewed". A service that needs to know someone has reached it is asking a yes or no question. A date of birth answers that question, and also says exactly how old the person is, for as long as the service keeps it.

Eligibility works the same way. A service that needs to know a household qualifies for support rarely needs the circumstances that make it qualify. Proportionate sharing means passing the answer and leaving the underlying record where it is.

Two ways Essence passes less

The first is selective disclosure. Essence uses SD-JWT, in which each claim in a credential is a separate item and the issuer signs only a fingerprint of each. The holder reveals the items a service asks for. The service checks each one against its fingerprint and checks the issuer's signature, so it knows the issuer vouched for what it saw and learns nothing about the rest. We built this in our August release. In testing, a holder over-shared a field nobody had asked for, and the verifier did not pass it on.

Selective disclosure also carries a ready-made answer. When an issuer puts one into the credential as a claim of its own, the kind of attribute the standard's "Person is Under 18" example describes, the holder reveals that claim alone. The service sees the answer and never the date.

The second is a predicate check in the platform's credentials service, in place since June 2026. The platform reads the value, works out whether the condition holds and, if it does, returns that answer with a signed commitment to the value. The verifier learns the answer, not the value, and because the answer comes signed it can check that nothing has been altered. The platform sees the value, because it does the working out; the service that asked does not.

Both are set out in more detail in Selective disclosure and predicate proofs (essence.digital/insights/selective-disclosure-and-predicate-proofs).

Why it matters most for people in vulnerable circumstances

The data models explain why support depends on sharing: "Often, no single organisation has all the data that could indicate the risk of a vulnerability, and no single organisation has all the support offers that are relevant". When support crosses organisations, a fact about a person passes through several systems. If what passes is an answer, each holds only what its step needed.

The person stays in view as well. Every field shared from an Essence vault has its own consent record, with terms the person sets and a withdrawal they control field by field; How consent works at Essence (essence.digital/knowledge/how-consent-works) explains how. What may be shared, and on what basis, remains a decision for each organisation and the law that governs it.

What this means for public services

When you design a check, ask only for what answers the service's question. An answer is better than a field, and a field is better than a document, because each step down is less to hold, secure and eventually delete. For any check you are offered, ask who sees the underlying value and what the service keeps. If you are working on eligibility or support services and want to see how these fit, we would be glad to talk.

Want to see it working? Request information

Published 10 October 2026

Related

When an AI agent asks for your data: delegation with limitsAn AI agent acting for a person should hold a narrow, expiring, countable slice of their consent, never their credentials. Delegated capabilities on Essence are built that way.
One set of sharing terms, many councilsCouncils on the same service write similar data-sharing agreements again and again. On Essence, the terms such an agreement sets are configuration, so one set written once can be applied by many organisations.