A polished website, a valid mobile driver's license, and even a valid reader authentication certificate are not proof that the organization asking for your data can be trusted.
A message arrives on a Tuesday morning: "Your account requires identity verification to continue."
The link opens a polished website. Your phone shows the same wallet prompt you have used at airport security and at the pharmacy. It asks whether you want to share your name, date of birth, and address from your mobile driver's license. The connection is encrypted. The credential is genuine. You tap Share.
The organization on the other side is not.
Nothing in that exchange was forged. No password was stolen. No cryptography was broken. The wallet did its job, the issuer's signature validated, and the presentation was delivered to the website that requested it. The failure was older and simpler: the person was persuaded to trust the wrong recipient.
We use the term Digital Identity Phishing for a phishing scenario in which an attacker persuades someone to authorize a genuine digital credential presentation to a deceptive service. The next evolution of phishing may not ask for a password. It may ask for a verified identity.
As digital IDs become more available for online transactions, this is a risk worth examining. The scale and pace of future exploitation remain uncertain. The underlying technical and trust questions, however, are already visible.
What is Digital Identity Phishing?
Traditional phishing targets secrets. The attacker wants a password, a card number, or a one-time code, and the fraud is exposed when the victim types the secret into a fake page. Digital identity phishing targets a decision. The attacker wants the victim to approve a legitimate wallet interaction with a recipient who should never have received the data.
That distinction changes the consequences. Unlike a password, personal identity information cannot simply be rotated after disclosure. Once exposed, its subsequent use and distribution may be difficult or impossible to control.
| Traditional credential phishing | Digital identity phishing | |
|---|---|---|
| What the attacker targets | A secret the victim knows | A disclosure the victim authorizes |
| What the victim sees | A fake login page | A genuine wallet prompt |
| What must be broken | A password, code, or session | Nothing cryptographic |
| Where the failure sits | The victim's credentials | The recipient's trustworthiness |
| What recovery looks like | Reset the secret | The data may already be disclosed |
What a digital ID and an mDL actually prove
A Digital ID is an electronic form of identification that a person stores on a phone or another device and presents to prove specific attributes about themselves. A Mobile Driver's License (mDL) is a familiar example: a digital version of a government-issued driver's license, built to the ISO/IEC 18013-5 standard and held in a compatible wallet.
Online presentation is defined separately. ISO/IEC TS 18013-7:2025 describes how mDLs and other mdoc credentials are presented to websites and remote services, including flows mediated by the W3C Digital Credentials API (DC API) and OpenID for Verifiable Presentations (OpenID4VP).
The privacy model matters. A well-designed wallet can support selective disclosure, allowing the holder to share only the attributes a transaction needs. A bar may verify age_over_21 without learning a birth date. A rental car counter may confirm a valid license without collecting a home address.
That is a genuine advance. It also raises a separate question: if the wallet can prove a fact without revealing everything, how does the holder know that the business asking for that fact has a legitimate reason to ask?
A wallet and its underlying protocols can help answer questions such as:
- Is this credential authentic and unaltered?
- Was it issued by a trusted issuer?
- Is the person presenting it the holder it was issued to?
- Is this presentation fresh and bound to this session?
- Which website or application is receiving the response?
Those checks do not, by themselves, establish whether the website belongs to a legitimate business, whether the business needs the requested data, or whether it will handle the data appropriately. Those are questions about the verifier: the organization requesting and consuming the identity information.
The attack does not need to break cryptography
Consider a possible scenario:
- A believable pretext. A fraudulent employer, lender, marketplace, or account-recovery page gives the victim a reason to complete verification quickly.
- A technically valid request. The service implements the relevant presentation flow, uses HTTPS, generates fresh nonces, and encrypts responses. None of that requires breaking a standard.
- A plausible attribute request. A job application might ask for a name, date of birth, address, or portrait. The request looks ordinary because similar information may be requested in legitimate transactions.
- A genuine consent interaction. The victim sees a wallet prompt and approves the disclosure.
- Misuse of the disclosed data. The recipient may use or distribute the information for fraudulent account opening, identity fraud, or targeted social engineering.
This is a threat scenario, not a claim that this exact sequence is occurring at scale. Its significance is that the cryptographic exchange can work as designed while the recipient is deceptive.
Why a valid signature is not a trust decision
Security teams often describe digital identity as a chain of cryptographic checks. Each check matters, but each answers a specific question. Confusing those questions is where the gap opens.
| Security control | Question it answers | Question it leaves open |
|---|---|---|
| Issuer signature | Is the credential authentic and unaltered? | Does the verifier deserve the data? |
| Holder binding | Is the presenter the legitimate holder? | Is the requester a legitimate business? |
| Session encryption | Is the exchange confidential in transit? | What happens after decryption? |
| Nonce and replay protection | Is this presentation fresh? | Can the recipient reuse the disclosed data? |
| Origin binding | Which origin is making the request? | Who operates that origin, and are they authorized? |
| Reader authentication | Does the verifier control a trusted key? | Who vetted that key, for what scope, and is it still valid? |
| Selective disclosure | Is the disclosure limited? | Is the verifier entitled to ask for those attributes? |
What origin binding does well
Origin binding is one of the important protections in online digital identity, and it is sometimes misunderstood.
When a presentation is mediated by the DC API, the browser or operating system can provide the wallet with the origin making the request, rather than relying only on information supplied by the page. ISO/IEC TS 18013-7 Annex C describes use of the origin received from the API in the session transcript and the requirement to abort if the API does not provide one. For signed OpenID4VP requests over the DC API, expected_origins is required, and the wallet compares the platform-provided origin against it. The response audience is also bound to that origin.
Implemented and validated correctly, these checks help defend against request relay and response replay. They make it harder to take a request intended for a legitimate service, forward it to a victim's wallet, and route the answer elsewhere.
But origin binding answers where a request came from. It does not independently answer who operates that origin or why the operator wants the data. A deceptive service may operate its own domain and ask for information directly.
What reader authentication does well
In ISO mDL deployments, reader authentication lets a verifier prove possession of a key and authenticate its request to the wallet. It is optional in ISO/IEC TS 18013-7, although an mDL may require it before releasing certain data elements online. OpenID4VP deployments can use signed requests and supported verifier-authentication mechanisms for related purposes, although the protocol details differ.
A signature proves possession of a key. The broader trust question depends on how that key became associated with an organization, what the organization was assessed against, and whether the organization remains authorized for the request it is making.
The standards recognize boundaries here. ISO/IEC 18013-5:2021 does not mandate methods for establishing trust in the certificates used as reader-authentication trust anchors; the way a reader CA is trusted by an mDL is outside the document's scope. The draft second edition retains a boundary around trust establishment and the wallet's use of reader information after authentication.
The point is not that reader authentication is ineffective. It provides an important technical assurance. The point is that authenticating a key and deciding whether an organization should receive personal information are different decisions.
A conformant verifier can still be deceptive
Imagine a fraudulent employment site that uses HTTPS, generates fresh nonces, supports the required encryption, and processes presentations correctly. Its protocol implementation could pass conformance tests. Its website could still make false promises about remote work, visa sponsorship, or account recovery to persuade people to disclose their identity.
Where the applicable wallet and ecosystem policies permit the request, the user may see a normal consent screen. If they approve, the service receives the attributes requested and disclosed.
Replay protections do not make disclosed information unusable. The recipient still possesses the data. It may be used in other fraud or social-engineering attempts.
Conformance establishes whether an implementation follows defined technical rules. It does not establish that the operator is honest, that its business purpose is legitimate, or that every requested attribute is justified. Those are different tests.
Two distinct ways trust can fail
A deceptive verifier does not have to enter an ecosystem through only one path.
Fraudulent enrollment is one possibility. Depending on the ecosystem's onboarding and certification requirements, a fraudulent organization may be able to obtain technically valid credentials. The strength of those requirements varies; it would be inaccurate to assume every trust framework is equally permissive.
Post-enrollment compromise or abuse is a different possibility. An organization may be legitimate when it is approved and later be compromised, change its business practices, or misuse access. A sound initial enrollment decision cannot guarantee future behavior.
Both scenarios raise questions that cryptographic validation alone cannot settle: what was established about the organization, what authority was granted, and what happens when circumstances change?
The intermediary problem
The problem becomes more complicated when a verifier service provider or verification aggregator acts on behalf of many downstream organizations.
A platform may be legitimate and properly vetted while serving hundreds or thousands of merchants, employers, or service providers. The platform's identity does not necessarily tell the holder which downstream business is requesting the information or whether that business has a legitimate purpose.
This creates several risks:
- Limited downstream visibility. An intermediary may know its customers, while wallets and other ecosystem participants see only the platform.
- Tenant compromise. An attacker who gains control of a downstream account may be able to operate through the platform's established integration.
- Uneven accountability. The platform may be responsible for a tenant's activity, but the ecosystem may lack clear attribution.
- Disproportionate suspension. Disabling a shared platform credential can disrupt legitimate customers along with the abusive tenant.
- Re-entry. A suspended business may attempt to return through a new identity or another intermediary.
These are structural risks of multi-tenant systems. They do not mean every intermediary is unsafe, or that every large platform has weak controls. They mean that platform-level identity alone may not tell the whole story.
Why revocation is not the whole answer
Revoking a reader-authentication certificate can give wallets a reason to reject future requests signed with that certificate. It is an important control, but it has limits.
First, revocation is forward-looking. It cannot retrieve information already disclosed.
Second, it depends on implementation and enforcement. Wallets must recognize the relevant trust framework, obtain sufficiently current status information, and apply a policy that rejects the request.
Third, one certificate is not necessarily an entire organization. A multi-tenant service may involve multiple keys, services, customers, and locations.
Finally, attackers may attempt to return under new identities. Suspension without meaningful accountability can become an administrative cycle.
The broader issue is not whether certificates should be revoked. It is whether the ecosystem can identify the relevant party, understand the scope of the problem, and respond proportionately.
What the standards and public research say
This is not a criticism of cryptography or of the standards process. It is a question about the operational environment in which standards are implemented.
NIST's National Cybersecurity Center of Excellence addresses the issue in NIST SP 1800-42A, its March 2026 initial public draft on mDLs for financial institutions. The guide identifies Malicious Verifier as a distinct threat: an attacker creates a legitimate-looking verifier, persuades users to present an mDL, and harvests personal information for later fraud or identity theft. The draft discusses user education, clear communication, the ability to cancel before presentation, and the limited availability of technical mitigations where verifier registration and certification programs are limited.
The protocol specifications also recognize verifier authorization as a relevant concern. The W3C Digital Credentials API discusses verifier authorization in its privacy considerations. OpenID4VP 1.0 defines verifier_info, which can carry attestations about verifier metadata, policies, trust status, or authorizations.
These mechanisms are building blocks. Their practical effectiveness depends on implementation, adoption, trust-framework governance, and wallet enforcement. A specification can define how information is carried and checked; it does not, by itself, establish the organizations and policies behind that information.
The privacy question
Any response to verifier trust must also respect privacy.
A system designed to establish whether a verifier may ask for information should not automatically become a system that records who presented what, when, and to whom. Those are different functions, with different privacy implications.
Jurisdictions and credential ecosystems will have different participation rules, disclosure policies, and governance models. Europe's EUDI framework illustrates one approach through relying-party registration and access certificates, including mechanisms described in Commission Implementing Regulation (EU) 2025/848. Other ecosystems may make different architectural choices.
The challenge is to improve confidence in the recipient without creating unnecessary tracking of identity presentations or assuming that one governance model will fit every jurisdiction.
What holders can do today
Digital identity phishing does not require a technical mistake, so user judgment still matters. Until wallets provide stronger and more consistent trust signals, a few habits can help:
- Check the organization shown in the wallet. A polished page and a padlock are not proof of legitimacy.
- Treat urgency as a warning. Job offers, refunds, account closures, and benefit deadlines can be used as pretexts.
- Match the request to the service. An age check generally does not require a home address. A newsletter signup does not need a portrait.
- Be cautious when the requester is unclear. If you cannot establish who is asking and why, pause before approving.
- Report suspicious requests. Wallet providers, issuers, and ecosystem operators need reports to identify patterns that individual users may not see.
For organizations requesting identity information, the corresponding responsibility is to explain why the information is needed, request only what is necessary, and make the recipient and purpose clear.
The questions the ecosystem still has to answer
Before sensitive identity information is released, the ecosystem needs credible ways to address three questions:
- Who is asking?
- On whose behalf are they asking?
- What establishes that the request is appropriate?
Digital identity has made meaningful progress in credential authenticity, privacy, and interoperability. The recipient side of the exchange deserves the same scrutiny.
The growth of digital identity phishing and the scale of future exploitation remain uncertain. We may be wrong about how quickly these risks develop or how widely they are exploited. But the distinction between proving a credential and establishing trust in its recipient is real today.
The future of digital identity security is not only proving who you are. It is also understanding who is asking for that proof.
Common questions
Is a mobile driver's license safe to use online?
An mDL can be used safely online when the wallet, presentation protocol, and recipient's trust mechanisms are implemented appropriately. The credential's cryptographic protections are important, but they do not independently establish the recipient's legitimacy.
Can a scammer steal my digital ID?
Digital identity phishing does not require stealing the credential or its keys. An attacker may persuade the holder to authorize a valid presentation to a deceptive service. The wallet and credential can work as designed while the data reaches the wrong recipient.
Does encryption prevent digital identity phishing?
No. Encryption protects data in transit. It does not establish that the recipient is legitimate, that the request is appropriate, or that the data will not be misused after delivery.
Is this attack happening at scale today?
NIST SP 1800-42A identifies malicious verifiers as a distinct threat in its threat model. That documents the risk, not widespread exploitation. The scale and pace of future digital identity phishing remain uncertain.
Technical references
- W3C, Digital Credentials Working Draft, including its privacy considerations and discussion of verifier authorization.
- OpenID Foundation, OpenID for Verifiable Presentations 1.0, including Appendix A on OpenID4VP over the Digital Credentials API and Section 5.11 on
verifier_info. - ISO/IEC TS 18013-7:2025, Personal identification: ISO-compliant driving licence, Part 7: Mobile driving licence (mDL) add-on functions, including Annex C on Digital Credentials API retrieval.
- ISO/IEC 18013-5:2021, Personal identification: ISO-compliant driving licence, Part 5: Mobile driving licence (mDL) application, including Annex B on certificate and CRL profiles.
- Draft ISO/IEC 18013-5 second edition. This is a draft, not a published standard.
- NIST SP 1800-42A, Digital Identities: Mobile Driver's License (mDL), Initial Public Draft, March 2026, especially the malicious verifier threat in Section 7.2 and trust establishment discussion in Section 10.1.
- Commission Implementing Regulation (EU) 2025/848, including Articles 7 and 8 on wallet-relying-party access and registration certificates.
The growth of digital identity phishing and the scale of future exploitation reflect ONEPROOF's current assessment of where the industry may be heading. These are informed assessments based on observable technical gaps and emerging trends, not established outcomes. We may be wrong about their scale or pace.


