A verifiable credential is a digitally signed claim from an issuer, held by the user and checked by a verifier without calling the issuer. A government can sign "this person is over 18", a bank can sign "this customer passed KYC" (know your customer checks), and the holder keeps both in a digital wallet. For a bank acting as the verifier, the check has three parts: the signature, the issuer, and whether the credential has been revoked. A fourth part sits outside the credential itself: the record the bank keeps of the check.
That last part is where most explainers stop and where compliance teams start. This guide covers what verifiable credentials are, how a bank checks one step by step, and what the standards leave for the bank to handle.
Issuer, holder and verifier: how verifiable credentials work
The W3C Verifiable Credentials Data Model 2.0, a W3C Recommendation since May 2025, describes "a three-party ecosystem" of issuers, holders and verifiers. The issuer makes a claim about someone and signs it. The holder receives the credential and stores it in a wallet. The verifier asks for proof of a claim and checks it.
The change from older identity systems is that the issuer and the holder are no longer the same party. In a federated login (signing in with a Google or Microsoft account, for example), the identity provider sits in the middle of every check: each time you sign in with a provider, it learns where you went. With a verifiable credential, the issuer signs once and steps out. The holder decides where to present it, and the verifier checks the issuer's signature on its own.
Two more ideas make credentials useful for regulated firms. Selective disclosure, which the W3C defines as "the ability of a holder to make fine-grained decisions about what information to share", lets a customer prove one fact without handing over the whole document. And decentralized identifiers (DIDs) give issuers and holders identifiers that resolve to public keys, so a verifier knows which key to check a signature against. Our guide to
decentralized identity covers DIDs in more depth.
How a bank checks a verifiable credential in five steps
Take a bank onboarding a customer who holds a KYC credential issued by another regulated firm. The check runs in five steps, and each one can fail in its own way.
- The issuer signs the credential. A trusted party, such as a government agency, another bank or a licensed KYC provider, checks the person once and signs a credential with its private key. What can fail: the issuer's own checks were weak, or its key was stolen.
- The holder stores it in a wallet. The customer keeps the credential on their device or in a hosted wallet. What can fail: the credential is presented by someone other than its holder, which is why many flows bind the credential to a key only the holder controls.
- The verifier asks only for what it needs. The bank requests specific claims, for example "is a resident of the EU" and "passed KYC in the last 12 months", not a copy of the passport. What can fail: the bank asks for more than its purpose needs, which data protection rules and eIDAS 2, the EU digital identity law covered below, both push against.
- The verifier checks signature, issuer and status. The bank confirms the signature matches the issuer's public key, confirms that issuer is one it accepts for this claim, and checks that the credential has not been revoked. What can fail: the status list is out of date, or checking it tells the issuer which bank asked and when.
- The verifier keeps a record of the check. The bank stores what it asked for, what it received, which issuer and status it checked, and when. What can fail: the record is incomplete, or it holds more personal data than the bank needed in the first place.
Steps 4 and 5 are where a bank's own controls take over from the standard.
Steps 1 to 3 are well covered by the standards.
Verifiable credentials compared with scanned documents and federated login
Banks already check identity today, mostly through document uploads or through logins from a third party. The table compares the three on the points a compliance lead cares about.
| Scanned document | Federated login | Verifiable credential |
|---|
| Who vouches | The bank's own review of the document | The identity provider at each login | The issuer, through its signature |
| What the verifier sees | The whole document | What the provider shares | Only the claims requested |
| How it is revoked | It is not; the copy stays valid on file | The provider disables the account | The issuer updates a status list or registry |
| What record is left | A copy of the document | Provider logs the bank does not hold | The bank's own record of the check |
The credential gives the bank less data to store and a check it can run without calling anyone.
Why verifiable credentials matter for banks in 2026
The European Union is the clearest driver. eIDAS 2 (Regulation (EU) 2024/1183), in force since May 2024, creates the European Digital Identity Wallet, where people hold their ID and "electronic attestations of attributes" such as proof of age or a diploma. For banks the key date is acceptance. Under Article 5f, private firms that rely on the wallet and must use strong user authentication (a high-assurance login rule), banks included, have to accept it no later than 36 months after the related implementing acts enter into force, when the user asks to use it. Worldline, a payments firm, puts Member State wallets at December 2026 and bank acceptance at December 2027 in its blog "Digital Identity and how banks can prepare for eIDAS 2.0 regulation" (February 2026).
The law also sets rules for how a bank uses the wallet. Article 5b(3) says relying parties, the firms that accept the wallet, "shall not request users to provide any data other than that indicated" when they registered, and Article 5a(5)(b) keeps issuers from learning how their attestations are used. That second rule matters for step 4: a revocation check that tells the issuer which bank asked would cut against it.
The US is moving on a slower track. NIST SP 800-63-4, the federal digital identity guidelines finalised in 2025, adds wallets that users control to its model for sharing identity between organisations. It binds federal agencies, not banks, but it is widely used as a reference.
What the standards leave to the bank
Two gaps stay open after a credential checks out.
The first is truth. The W3C model says it directly: "Verifiability of a credential does not imply the truth of claims encoded therein." A valid signature shows who made the claim and that it was not changed. Whether the issuer checked well is still a question of which issuers the bank trusts, and for which claims.
The second is the record. The Financial Action Task Force (FATF), which sets the global anti-money laundering standards that banks follow, says in its Guidance on Digital Identity (March 2020) that "the requirements of Recommendation 11 (Record-keeping) apply equally to recordkeeping in digital" form, and that "the ultimate regulatory responsibility for CDD measures remains with the regulated entity". CDD is customer due diligence: a bank that accepts a credential still owns it and still has to show an examiner what it checked. A credential that minimises data on the way in does not remove the need for a record on the way out. Our post on
AML and KYC without mass surveillance looks at how firms keep that record without building a new data store of personal files, and the
Revolut KYC data breach shows the cost when the data store is the weak point.
Where zkDatabase fits in a verifiable credential system
zkDatabase, the
verifiable database built by Orochi Network, is designed to cover steps 4 and 5 for banks and the platforms they work with.
A Zero-Knowledge Proof lets one party show that a statement is true without showing the data behind it. Here, it lets a customer show "my credential is on the valid list" without showing which one.
For status, zkDatabase is built to act as a verifiable credential registry. Issuers write valid credentials into the database, each update comes with a Zero-Knowledge Proof that it was made correctly, and the set of valid credentials is committed as a Merkle root: a short fingerprint of the whole list, published so that any change to the list shows up. When a credential is revoked, a new root is published, and a proof against the old root no longer passes. A holder then proves membership in the valid set with a Zero-Knowledge Proof, without revealing which credential they hold. The bank checks that proof against the current root, so it confirms the credential is still valid without asking the issuer, and the issuer does not learn who checked.
For the record, zkDatabase keeps the history of credential set states with Merkle proofs. A compliance team can show which credentials were valid at the time of a check, and an auditor can confirm it without access to the personal data behind them. Many tokenized asset platforms check eligibility today with an allow-list, where a registry operator states that a wallet is approved. The registry replaces that statement with a proof that a credential is in the valid set.
The proof shows that the registry was updated and queried as claimed. It does not show that the issuer's original KYC was good: that stays with the issuer and with the bank's choice of which issuers to accept. For a wider view of how identity proofs fit financial institutions, see our guide to
zero-knowledge KYC and decentralized identity and our post on
reusable KYC.
Frequently asked questions
Are verifiable credentials the same as W3C verifiable credentials?
W3C verifiable credentials are the most widely used format: the W3C Verifiable Credentials Data Model 2.0 sets the structure. The EU wallet also uses other formats, such as SD-JWT VC and ISO mdoc, which follow the same issuer, holder and verifier model.
How does verifiable credential verification work without calling the issuer?
The verifier checks the issuer's digital signature against the issuer's public key, which it gets from a DID document or a trusted list. It then checks a status list or registry for revocation. Neither step needs a live call to the issuer.
Can a bank rely on a credential issued by another bank?
It can accept the credential as evidence, but FATF guidance keeps the responsibility for customer due diligence with the bank that relies on it. The bank decides which issuers it trusts for which claims, and keeps its own record of the check.
Explore zkDatabase
See how zkDatabase proves credential status and keeps a check record auditors can verify without seeing personal data.