Join our Newsletter — 33% off our NHI Course
Home› Guides› Digital Identity Wallet Fraud Risks Guide: Attacks, Controls…
Guide Customer Identity

Digital Identity Wallet Fraud Risks Guide: Attacks, Controls and Detection Signals

PDF for newsletter subscribers · free to read online

← All guides
By NHI Mgmt Group, edited by Lalit Choda Updated 27 October 2026 14 min read
On this page

A digital identity wallet makes credentials hard to forge, but it does not stop fraudsters from getting a genuine credential issued to the wrong person, taking over the wallet that holds it, or tricking its owner into presenting it to them. As wallets such as the EU Digital Identity (EUDI) Wallet and mobile driving licences move into banking, telecoms and public services, the fraud shifts from fake documents to the seams around the wallet: onboarding, device recovery, presentation requests and the relying party's own checks. This guide maps those attack paths, the controls the EUDI Architecture and Reference Framework (ARF) and the OpenID protocols provide, the gaps that remain for relying parties and wallet providers, and how to detect and measure fraud. It is for fraud, identity and security teams at relying parties, wallet providers and credential issuers. It builds on the Digital Identity Wallets Guide and does not repeat the liveness and document checks covered in the Deepfake-Resistant Identity Verification Guide.

Key takeaways

  • Fraud moves to issuance. A wallet credential is only as good as the identity proofing behind it; the ARF requires PID issuance to meet Level of Assurance High under Implementing Regulation (EU) 2015/1502, but other attestations only need security commensurate with their use.
  • Wallet takeover is account takeover. Lost devices, weak recovery and re-activation on a new phone are the routes to someone else's credentials, so revocation and activation controls matter as much as cryptography.
  • Presentation phishing is real. A fake site can relay a genuine relying party request, especially through cross-device QR flows, so the victim presents a credential that completes the attacker's session.
  • Relying parties must verify, not assume. The ARF says verifying the device binding signature is recommended rather than mandatory for relying parties, and SD-JWT VC device binding is optional; skipping these checks is a choice to accept copied credentials.
  • Privacy limits signals. The ARF rules out behavioural data, geolocation and active phone call detection inside the wallet for privacy reasons, so relying parties must build fraud detection from their own session data.

The wallet fraud threat model

The EUDI risk register, set out in Annex I of Implementing Regulation (EU) 2024/2981 and summarised in ARF chapter 7, lists high-level risks that map directly to fraud: creation or use of a fake electronic identity (R2), fake attributes (R3), identity theft (R4), unauthorised transactions (R9) and transaction manipulation (R10). Its technical threats include phishing and spoofing (TT5.2), replay of messages (TT5.3), brute force (TT5.4) and malware (TT5.7). The table groups these into the five attack paths fraud teams see in practice.

Attack pathWhat the fraudster wantsWhere it happensMain owner of the control
Onboarding fraudA genuine credential issued for a stolen or synthetic identityPID or attestation issuance, wallet activationIssuer and wallet provider
Wallet takeoverControl of a victim's wallet and its credentialsDevice theft, re-activation, recovery, malwareWallet provider, user
Presentation phishingThe victim presents a credential into the attacker's sessionRemote flows, especially cross-device QRRelying party, wallet
Relay and replayReuse of a presentation or a live relay to a distant verifierProximity (NFC, BLE) and remote flowsRelying party
Rogue or over-asking relying partyHarvest of personal data for later fraudPresentation requests from unregistered or misrepresented servicesRegistrar, wallet, user

Onboarding fraud: getting a real credential for the wrong person

Cryptographically signed credentials make forgery impractical, so the attacker's best option is to pass the issuer's checks. The ARF leaves the identification method to each provider but sets the bar: for a PID the processes "will satisfy the requirements for Level of Assurance High" in Implementing Regulation (EU) 2015/1502; for a qualified attestation they follow the reference standards in Implementing Regulation (EU) 2025/1566; other attestations use a level "commensurate with the required level of security".

  • Weak-link issuance: a high-assurance PID can be used to obtain lower-assurance attestations, and those attestations can then be presented where the relying party does not check which issuer or assurance level stands behind them. Relying parties should accept only attestation types and issuers that match the risk of the transaction.
  • Activation on the attacker's device: the ARF describes ways to ensure a new PID "can only end up on the device used by the subject of the PID": user authentication to the wallet, mutual authentication between wallet and PID provider under OpenID4VCI, and authentication to the PID provider with an eID means at LoA High (or LoA Substantial plus additional remote onboarding under Article 5a(24) of the European Digital Identity Regulation). Alternatives are identity matching inside the WSCA and an activation code sent over a trusted channel.
  • Remote proofing attacks: where an issuer relies on remote document and selfie checks, the deepfake, injection and presentation attacks covered in the Deepfake-Resistant Identity Verification Guide apply in full. See also the Identity Proofing and KYC Guide.
  • Unauthorised issuers: an attestation provider may try to issue a type of attestation it is not authorised for, such as a diploma. Wallets check the provider's registration certificate, and the ARF notes that relying parties can query the Registrar's online service to confirm an issuer is registered for that attestation type.

Wallet takeover: device theft, re-activation and recovery

The ARF requires two layers of user authentication. During activation the wallet forces the device operating system to enable multi-factor authentication (possession of the device plus knowledge or inherence). A second mechanism in the WSCA/WSCD (the secure application and secure hardware that hold keys) is used whenever a PID or a WSCD-bound attestation is issued, presented or deleted. That makes casual theft hard. Takeover therefore tends to come through the processes around the device.

Takeover routeControl in the ARFWhat relying parties and providers should add
Stolen unlocked phoneWSCD authentication for PID presentation; user can ask the wallet provider to revoke the Wallet Unit after loss or theftFast, well-advertised revocation channel; relying parties check revocation status for high-value actions
Re-activating a victim's PID on a new deviceActivation options tied to eID authentication, WSCA identity matching or an activation code over a trusted channelTreat wallet migration as a high-risk event: notify the old device and the user's contact channels
Social engineering of the wallet provider's support or recovery processOut of scope of the ARFRecovery that re-proofs identity, not support staff discretion. See the Account Recovery and Help Desk Security Guide
Rooted, emulated or malware-infected deviceWallet provider monitors posture signals such as rooting, jailbreak, emulators and malware, with a four-level response up to revoking the Wallet InstanceRelying parties honour the revocation status in the wallet instance attestation where available
SIM swap or email takeover used to reset accountsNot addressed by the wallet itselfDo not let a phone number or email alone re-bind a wallet. See the SIM Swap Fraud Guide

The ARF's optional four-level posture framework is a useful template even outside the EUDI ecosystem: level 4 (confirmed root or jailbreak, debugger or emulator, compromised key, failed integrity check) revokes the Wallet Instance and its keystores and forces reinstallation on a clean device; level 3 (unpatched high-risk OS, failed biometric attempts above a high threshold) revokes and requires step-up or re-activation; level 2 limits high-value presentations; level 1 allows full use. Recovery is where most takeovers start: the 2026 Meta AI Instagram account takeover worked because an AI support assistant sent reset links to attacker-supplied email addresses without checking them, and accounts with two-factor authentication were not taken over.

Presentation phishing and cross-device relays

In a remote presentation the relying party sends a request and the wallet returns a signed response. If the attacker can place a genuine request in front of the victim, the victim's wallet will produce a genuine response for the attacker's session. The common pattern is a fake website that starts a real session with the relying party, shows the victim the relying party's QR code, and waits for the victim to scan it with their wallet. RFC 10027 (BCP 247, August 2026) names this cross-device consent phishing and says implementers should avoid cross-device flows for sensitive actions if the risk cannot be mitigated.

  • Prefer same-device flows: HAIP 1.0 makes same-device support mandatory and, with OpenID4VP, binds the response to the session that made the request. The W3C Digital Credentials API lets the browser pass the request to the wallet and bind it to the calling website's origin.
  • Bind every response: OpenID4VP binds the presentation to the relying party's client identifier and a fresh nonce. Reject any response where the nonce or audience does not match, and do not accept presentations without holder binding for anything sensitive, because they offer no replay protection.
  • Show the user who is asking: the EUDI wallet authenticates the relying party instance from its access certificate and checks its registration certificate, warning the user when the request asks for more than the relying party registered. Relying parties should keep the name in their certificate consistent with their brand so users can recognise it.
  • Use transaction data for high-value actions: OpenID4VP's transaction_data lets the wallet sign over what the user is approving, for example a payment amount, so a relayed approval cannot be repurposed. The ARF notes that its semantics must be defined in the relevant rulebook or specification.
  • Close cross-device gaps: short request lifetimes, single-use requests and a check that the browser that started the session is the one that receives the result. See the Cross-Device Flow Security Guide and the QR Code Phishing (Quishing) Guide.

Relay, replay and credential sharing

  • Replay of a captured presentation: blocked by nonce and audience binding when the relying party checks them, and by device binding (called mdoc authentication in ISO/IEC 18013-5 and key binding in SD-JWT VC). The ARF makes device binding mandatory for PIDs and ISO/IEC 18013-5 attestations but only recommended for SD-JWT VC, and OpenID4VP lets the relying party demand it with require_cryptographic_holder_binding.
  • Live proximity relay: an attacker relays radio traffic between a victim's wallet and a distant reader so the presentation appears local. Supervised presentations with human inspection of the portrait reduce this, as does requiring user confirmation on the wallet for every presentation (the ARF requires user approval in all use cases).
  • Credential sharing or coercion: device binding proves the credential is on a genuine wallet, not that its owner is the person presenting. ARF section 6.6.3.9 describes user binding options: trusting the wallet's certified user authentication, comparing the portrait with the presenter (with presentation attack detection and injection or deepfake detection when remote), or comparing against a photo ID.
  • Wallet-to-wallet fraud: the ARF limits wallet-to-wallet presentations to proximity, and proof that the verifier's wallet is genuine will depend on a future common method. Treat person-to-person presentations, such as selling a car, as lower assurance until then.

Rogue relying parties and data harvesting

Fraud often starts with data collected for a different purpose. The ARF counters over-asking with three measures: selective disclosure under user control, mandatory relying party registration of the attributes it intends to request (published by the Member State Registrar and included in a registration certificate), and disclosure policies that issuers can embed in attestations. Relying parties found in violation have their access certificates revoked. The ARF also warns about linkability: unique values such as salts, hashes and public keys in attestations can be used to track users, so relying parties are expected to discard those values once they no longer need them.

  • Request the minimum attributes, for example an age-over attestation instead of a date of birth.
  • Do not store raw presentations longer than the transaction needs.
  • Monitor for lookalike services impersonating your brand that ask users to present credentials, and report them to the Registrar or competent authority.

Controls by role and decision guidance

DecisionLower-risk transactionHigher-risk transaction
Accepted credentialAny registered attestation of the right typePID or qualified attestation from a listed issuer
Device binding proofRequested where supportedRequired and verified (require_cryptographic_holder_binding true)
FlowSame-device or cross-deviceSame-device or Digital Credentials API; cross-device only with proximity checks
Revocation checkCached status listFresh status check at presentation time
User bindingTrust the wallet's user authenticationAdd portrait comparison with PAD or a supervised check
Transaction dataNot neededSigned transaction_data describing what is approved

Small relying parties using a verification intermediary should confirm in the contract that the intermediary verifies nonce, audience, device binding and revocation, because the ARF treats intermediaries as acting for the relying party. Large relying parties running their own verifier should test these checks with negative test cases in every release.

Detection and measurement

  • What to log: request identifiers, nonce, flow type (same-device, cross-device, Digital Credentials API), issuer and attestation type, device binding result, revocation result, time from request to response, and the IP and user agent of the session that started and finished the flow.
  • Signals of presentation phishing: cross-device responses where the starting session and the wallet's network are far apart, many requests created but not completed from one source, and very short gaps between request creation by an unusual client and a response.
  • Signals of takeover: presentations from a wallet instance shortly after the user reported loss, new wallet activations followed by immediate high-value use, and repeated failed presentations.
MetricWhy it matters
Percentage of presentations with verified device bindingShows exposure to copied or replayed credentials
Percentage of high-value flows on same-device or Digital Credentials APITracks reduction of cross-device phishing exposure
Responses rejected for nonce, audience or revocation failuresA rising count can indicate active attack or integration faults
Fraud losses per 10,000 wallet-based onboardings, compared with document-based onboardingShows whether the wallet is actually reducing fraud
Median time from reported device loss to revocationMeasures the takeover window

Framework mapping

Wallet fraud controlNIST CSF 2.0NIST SP 800-53 Rev 5ATT&CK v20
Issuance identity proofingPR.AA-02IA-12
Verify relying party and wallet authenticationPR.AA-03IA-8, IA-2
Session binding against relay and phishingPR.AA-03SC-23T1566 Phishing
Detect anomalous presentationsDE.AE-02SI-4T1111 MFA Interception

Common failure patterns

  • Accepting any signed credential: checking the signature but not the issuer's trust list entry, attestation type or assurance level, so a low-assurance attestation unlocks a high-risk service.
  • Skipping device binding: treating it as optional because the ARF says relying party verification is recommended, then accepting presentations that could be copied.
  • Cross-device QR for high-value actions: using a desktop QR flow for account opening or payments with no proximity check, the exact pattern RFC 10027 warns about.
  • Ignoring revocation: caching status lists for days, so a wallet reported stolen still works.
  • Recovery that bypasses the wallet: letting a user re-bind a wallet or account through email or a support call, which reopens every takeover route the wallet closed.
  • Keeping full presentations: storing raw credentials and their unique values, which creates the data store a later breach will expose.

Lessons from incidents

No wallet-specific fraud incident is yet documented in the NHI Mgmt Group breach database. The closest lesson is from recovery abuse: the Meta AI Instagram account takeover affected 20,225 people because an automated support process trusted an unverified email address. Wallet providers that automate support or recovery, including with AI assistants, must require the same proof as activation. See the Account Takeover Prevention Guide.

Practitioner checklist

  • Map each service to an accepted attestation type, issuer list and assurance level, and reject anything else.
  • Verify nonce, audience, signature chain, device binding and revocation on every presentation, and test each check with negative cases.
  • Move high-value flows to same-device presentation or the Digital Credentials API; add proximity checks if cross-device must stay.
  • Use signed transaction data for payments and other approvals that a relay could repurpose.
  • Agree a fast revocation route with wallet providers and publicise it to users who lose their phone.
  • Treat wallet activation on a new device and recovery as high-risk events with notification and step-up.
  • Request the minimum attributes and discard unique values after the transaction.
  • Log flow type, binding and revocation results, and track the five metrics above monthly.
  • Write wallet attack paths into fraud scenarios and red-team exercises, including cross-device phishing.

Standards and references

Related NHI Mgmt Group resources: Digital Identity Wallets Guide · Relying Party Integration with Digital Identity Wallets Guide · Wallet Unit Attestation and Device-Bound Keys Guide · Citizen Digital Wallet Safety Guide · EUDI Wallet Architecture and Reference Framework Guide

PDF for subscribers

Take this guide with you

Every section, table and checklist in one PDF, with links to the sources and related guides. Subscribe to our weekly NHI & AI Identity Journal and we’ll email you a download link.

Free weekly newsletter

Subscribe to the NHI & AI Identity Journal

The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.

Bonus 33% off our NHI Course when you subscribe.

NHIMG Editorial Note
Researched and written by the NHI Mgmt Group editorial team using AI tools, under the editorial direction of Lalit Choda. Facts are drawn from the sources listed on this page and checked by our automated quality process. Last updated 27 October 2026. Spotted an error? Use our contact page.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM and Agentic AI security. nhimg.org

PDF Downloads for Newsletter Subscribers

Get the guide

Enter your details and we’ll email you a download link.

The full guide is free to read online. The PDF is a thank you for subscribing.

Fill in all fields marked * and tick the PDF box to continue.