When you hand a bartender your driver's license to verify your age, they see your name, your photo, your address, and your exact date of birth. All of that information exists on the license, and presenting the license means presenting all of it. The bartender needs exactly one fact: whether you are 21 or older. Everything else is irrelevant to the transaction but disclosed anyway, because physical credentials do not support any other model.
Selective disclosure is the technical capability that changes this constraint. A credential system with selective disclosure lets the holder share only the specific attributes needed for a given transaction, while keeping the rest of the credential private. For digital identity systems built around privacy and least-privilege access, selective disclosure is not a convenience feature. It is a structural requirement.
What Selective Disclosure Actually Does
A verifiable digital credential typically contains a bundle of attributes: name, date of birth, address, license class, expiration date, issuance jurisdiction, and potentially more. In a traditional presentation, the credential is verified as a whole. Every attribute travels with the presentation, whether the verifier needs it or not.
Selective disclosure breaks that coupling. With the right cryptographic construction, a credential holder can present a single attribute (say, "age over 21") or a subset of attributes (say, "name and expiration date") while the rest of the credential remains hidden. Critically, the verifier can still confirm that the disclosed attributes came from a real, unmodified credential signed by a trusted issuer. The integrity check still works, even on a partial disclosure.
This is architecturally distinct from a credential that simply contains less information to begin with. An issuer issuing a credential with only one attribute solves a different problem. Selective disclosure operates on a rich, multi-attribute credential: the holder decides what to reveal at presentation time, within the bounds of what the issuer signed.
Three Approaches, Different Trade-Offs
Several technical approaches to selective disclosure are in active deployment or standardization. They share the same goal but arrive there through different cryptographic methods, and the differences matter for architects choosing a format.
SD-JWT (Selective Disclosure for JWTs) is the approach defined in IETF RFC 9901 (published November 2025) and the basis of the SD-JWT Verifiable digital credentials profile, which is progressing through IETF standardization as an Internet-Draft with publication expected later in 2026. In SD-JWT, each claim in the credential is individually hashed and salted. The issuer signs the full set of hashed claims. At presentation, the holder reveals only the plaintext values for the claims the verifier needs; all other claims remain as undisclosed hashes. The verifier can confirm the issuer's signature over the full set and confirm that the disclosed claims match their hashed representations, without learning anything about the undisclosed claims.
SD-JWT is relatively simple to implement and integrates with existing JWT tooling, which has contributed to its adoption in the EU Digital Identity Wallet Architecture and Reference Framework (EUDI ARF), where SD-JWT VC is one of two mandatory credential formats alongside mdoc (the format used by ISO/IEC 18013-5 mDL). The trade-off is that selective disclosure decisions are determined at presentation time by revealing or withholding hashes, but the structure of the credential (the set of claim types that exist) is visible to any party who holds the credential, even before disclosure.
mdoc selective disclosure follows a different model. The ISO/IEC 18013-5 standard, which governs mobile driving licenses (mDLs), specifies selective disclosure natively. The issuer signs a message authentication code over each data element independently using a device key, so at presentation the holder can prove individual elements without revealing others. The verifier requests specific data elements; the wallet responds only with those elements, each with its own integrity proof. The ISO/IEC 18013-7 standard, published as a technical specification in 2024 and updated in 2025, extends this to online presentations, including support for OpenID for Verifiable Presentations (OID4VP).
Zero-knowledge proofs (ZKPs) represent the most expressive approach and the most computationally intensive. Rather than revealing a claim value at all, ZKPs allow a holder to prove a predicate about a claim: "this person is over 21" without disclosing the actual birth date, or "this person's income is below a threshold" without disclosing the exact figure. Approaches such as BBS+ signatures, currently progressing through standardization as a W3C Candidate Recommendation for verifiable digital credential cryptosuites and as an IRTF Internet-Draft for the underlying signature scheme, can produce unlinkable proofs, meaning that multiple presentations of the same credential cannot be correlated by a colluding verifier or network observer, even when the same credential is used repeatedly.
ZKPs offer the strongest privacy properties but come with implementation complexity, limited tooling maturity, and ongoing standardization work. For most government deployments today, SD-JWT and mdoc selective disclosure are the practical choices; ZKP-based approaches are tracking closely behind.
Why This Matters for Data Minimization
Data minimization is the principle that a system should collect, process, and retain only the personal data necessary for a specific, defined purpose. It appears in the Privacy Act of 1974, in the NIST Privacy Framework (Version 1.0, with Version 1.1 in development as of 2025), and in state privacy statutes with increasing frequency. Selective disclosure is the technical mechanism that makes data minimization achievable at the attribute level during credential presentation.
Without selective disclosure, a verifier who needs only one fact about a person must receive a complete credential containing many facts. The verifier may or may not be disciplined about discarding the rest. The system creates a structural incentive to over-collect: because the technical model bundles everything together, the verifier sees everything. Compliance with minimization requirements then depends entirely on the verifier's data handling policies and practices, not on the architecture.
With selective disclosure, the architecture enforces the constraint. A verifier who requests only age-over-21 receives only that disclosure. The system does not make minimization a policy aspiration; it makes over-collection technically impossible within the presentation itself. This distinction matters for agencies that process large volumes of credential presentations and cannot rely solely on contractual or policy controls to prevent attribute accumulation.
The Proportionality Principle in Government Transactions
Government services span a wide range of verification requirements. Age verification for a hunting license is a different problem from income verification for benefits eligibility, which is different again from identity proofing for access to sensitive agency records. Each transaction has a specific set of facts it legitimately needs; the rest is outside the scope of that transaction.
Selective disclosure makes it possible to match the data presented to the actual verification requirement. A state agency verifying residency for voter registration does not need a person's name if the record-matching is done by credential identifier; it certainly does not need the person's medical information if that attribute happens to live on the same credential from a health agency. The principle that data collection should be proportionate to its purpose is longstanding in privacy law, but digital systems have historically made proportionality difficult to enforce technically.
Consider the case of a public benefits application that requires proof of age, proof of residency, and proof of income. Under a traditional model, the applicant might submit three separate documents, each revealing more than the program requires. Under a selective disclosure model, a single credential from the relevant issuing authority could satisfy all three requirements with exactly the attributes each program rule specifies. No additional attributes travel across the disclosure. The issuer's signature guarantees authenticity; the selective disclosure mechanism guarantees minimization.
This architecture also supports the kind of credential reuse that reduces burden on residents: a person who has already obtained a verifiable digital credential from one agency can present specific attributes from that credential to a second agency, without the second agency needing to contact the first or the resident needing to obtain a separate document.
Selective Disclosure and Zero Trust Identity Architecture
Zero trust security architecture operates on a principle of least-privilege access: no user, system, or network should receive broader access than the specific task requires, and every access request should be explicitly verified. Selective disclosure applies exactly this logic to identity attribute disclosure. Each transaction should receive the minimum set of attributes sufficient to make the required verification decision, and the architecture should enforce that limit rather than relying on policy compliance after the fact.
In a zero trust identity model, the credential presentation itself becomes a narrow, scoped assertion rather than a broad data handoff. The verifier requests what it needs; the wallet discloses only that. The issuer's signature provides the integrity and authenticity guarantee without the issuer being involved in or notified about the presentation (a property sometimes called "issuer unlinkability" or the absence of a "phone home" requirement). The holder retains control not just over whether to present a credential but over precisely what the presentation reveals.
This property matters for government systems beyond the privacy calculus. When a verifier receives only the attributes it needs, it also reduces its own exposure: it is not accumulating a detailed profile of the person, it is not retaining sensitive attributes that become a liability if the system is breached, and it is not creating a data record that must be managed, secured, and eventually purged. Selective disclosure reduces the attack surface by reducing what verifiers hold.
How Selective Disclosure Works with Presentation Requests
Selective disclosure does not operate in isolation; it is connected to the presentation request that a verifier sends to a wallet. The request specifies which attributes are needed for the transaction. A well-designed system lets the wallet confirm that the requested attributes match the verifier's stated purpose, and lets the holder review and approve the disclosure before anything is sent.
This creates a meaningful consent moment. The holder can see exactly what will be disclosed, to whom, and for what stated purpose, before confirming the presentation. Contrast this with a model in which a person hands over a physical document or uploads a scan: in neither case does the person have a clear picture of which attributes the receiving system will extract, store, or share.
The combination of presentation request specificity and selective disclosure technical enforcement is what makes verifiable digital credentials meaningfully different from prior digital document models. A PDF of a driver's license is a digital document but not a selectively disclosable credential. The format matters because the format determines what privacy properties are technically achievable.
Selective Disclosure in State Digital Identity Programs
Several states are building digital identity programs that incorporate selective disclosure as a design requirement rather than an afterthought. Utah's SB 275, which established the State-Endorsed Digital Identity (SEDI) framework (effective May 6, 2026), requires that credential presentations request only the minimum attributes necessary for a transaction and that holders be able to share only what a transaction requires, such as proving age eligibility without disclosing an exact birth date or home address. SEDI frames the digital identity system as serving the resident's interests, which implies that credential presentations should not expose more about the resident than the transaction requires.
For state architects designing systems under frameworks like SEDI, selective disclosure capability in the chosen credential format is not an optional enhancement. If the system issues credentials in a format that does not support selective disclosure (a flat signed JSON blob, for instance), no amount of policy writing will enable minimized presentations. The technical capability must be present in the format and the wallet implementation from the beginning; retrofitting it later requires re-issuance of credentials and potentially changes to the verifier integration layer.
Choosing a credential format is therefore a privacy architecture decision, not just a technical one. The choice between SD-JWT VC, ISO mdoc, and other formats carries direct implications for what selective disclosure properties are achievable in the system you build.
What Selective Disclosure Does Not Solve
Selective disclosure addresses the presentation layer: what attributes travel from the wallet to the verifier during a specific transaction. It does not govern what the verifier does with those attributes after receipt. A verifier who receives only an age-over-21 disclosure could still log that transaction in a way that builds a behavioral profile over time. Selective disclosure removes the structural incentive to over-collect at the moment of presentation; it does not eliminate every downstream privacy risk.
Similarly, selective disclosure does not address issuer-side data minimization. If an issuing agency includes unnecessary attributes in a credential at issuance, selective disclosure ensures those attributes need not be shared at presentation, but the attributes still exist in the credential and on the issuer's records. Proper privacy architecture requires minimization at both the issuance and presentation stages.
There is also the question of correlation. Even with selective disclosure, repeated presentations of the same credential to the same verifier can be correlated by the credential identifier if that identifier is stable and disclosed. Some selective disclosure schemes address this with unlinkable proof techniques (as in ZKP-based approaches), while others leave identifier correlation as a residual risk. Architects who need to prevent verifier-side profiling need to evaluate these properties at the scheme level, not assume that selective disclosure alone provides unlinkability.
Building for Selective Disclosure from the Start
The practical implication for architects and program leads is that selective disclosure capability must be treated as a first-order design requirement, not a feature to add later. This means selecting credential formats that support attribute-level disclosure, ensuring the wallet implementation exposes the selective disclosure surface to the holder during presentation, designing the presentation request layer to request only what is needed, and establishing governance rules about what verifiers may request.
It also means considering the holder's experience: selective disclosure should not create a burden where users must manually curate attribute lists for every transaction. Well-designed systems encode the verifier's legitimate purpose in the presentation request and surface a clear, confirmable disclosure summary to the holder. The technical capability and the user experience must be designed together.
Selective disclosure does not make privacy automatic, but it makes privacy-preserving credential architecture possible. Systems that do not include it at the foundation will face a structural ceiling on the privacy properties they can achieve, regardless of other controls applied on top.
To understand how credential formats affect what privacy properties are technically achievable, read What Is a Credential Format, and Why Does It Matter Which One You Choose? For a broader look at verifiable digital credentials as infrastructure, see How Do Verifiable Digital Credentials Work?
About SpruceID: SpruceID builds digital trust infrastructure for government. We help states and cities modernize identity, security, and service delivery — from digital wallets and SSO to fraud prevention and workflow optimization. Our standards-based technology and public-sector expertise ensure every project advances a more secure, interoperable, and citizen-centric digital future.