Research program 04
VEID, the Verifiable Electronic Identity layer of VirtEngine, is decentralised protocol technology: multi-factor verification whose evidence is processed and encrypted on the user's own device, scored by chain consensus rather than a central bureau, and presented to services as proofs of attributes — not copies of documents.
01The pipeline
Verification strength usually comes at privacy's expense: stronger checks, bigger central databases. VEID's pipeline is engineered to break that trade — evidence is captured and encrypted on-device, validators score it under consensus, and only the score becomes ledger state.
The reference capture flow combines document capture with guided framing and OCR field extraction; selfie capture with active liveness challenges — blink, head turn, smile — and anti-spoofing signals; fingerprint or iris capture through the platform's secure biometric APIs; and device integrity attestation via Google Play Integrity or Apple App Attest, with SafetyNet and DeviceCheck fallbacks for legacy devices.
Biometric templates and attestation payloads are encrypted into the VEID envelope before anything leaves the device, and stored as consent-gated identity scopes — including dedicated scope types for biometric hardware and device attestation. Data minimisation is a design rule: only the templates and integrity metadata required for verification are stored, encrypted, with lifecycle expiration and revocation.
Block-proposing validators decrypt submitted scopes with their private keys, score the evidence with machine-learning models, and write the resulting identity trust score into the proposed block; consensus validators repeat the computation and vote. No single verifier's judgement is final, and no central database of documents accumulates anywhere.
The x/veid/zk package verifies proofs about identity attributes — validity, tier, uniqueness, or a predicate like being over 18 — without placing the attributes themselves on-chain or handing documents to the relying service. Services learn the answer to their question and nothing else.
02Verification tiers
The chain derives a tier from the identity trust score. Services ask for the tier their risk actually requires — a real-person signal for a forum is not the same question as validator onboarding.
| Tier | Evidence | Unlocks |
|---|---|---|
| 0 Unverified | none — where every wallet starts | browsing; activity that needs no identity checks |
| 1 Basic | self-entered details; a session passing the network's early checks | low-risk services needing a real-person signal |
| 2 Standard | government document (front and back) + selfie with active liveness + OCR match | most participating services; age and attribute proofs |
| 3 Trusted | Standard plus hardware-backed biometrics, device attestation, sustained history | high-trust roles — e.g. validator onboarding; high-value transactions |
Attestation-required policies apply to high-risk actions: where a device cannot support hardware attestation, the flow records that fact and verification pipelines restrict what the result may be used for.
03The service
The Foundation stewards identity.org.au, the plain-language public website for individuals and relying services. The wallet portal is planned for the January 2027 launch window; VEID itself is decentralised validator-operated protocol technology.
identity.org.au explains the wallet, the verification levels, consent and revocation, and what relying services can and cannot see — written for citizens, not cryptographers. Join the launch list for the planned wallet portal. identity.org.au is a community service stewarded by the not-for-profit foundation, and it is clearly labelled as not an Australian Government service.
The wallet is a reference implementation: the open repository carries the React Native capture application (mobile/veid-capture-app) implementing the document, selfie, liveness, biometric-hardware, and attestation flows, with encryption and native ML modules as documented, pluggable extension points. We make no app-store claims and publish no user numbers — the verifiable artifacts are the code, the design documents, and the accepted patent.
04Rights and consent
The identity program is the most direct expression of the foundation's third purpose: promoting privacy, dignity, autonomy, and equality of access — and protection from unlawful surveillance and private capture of essential digital infrastructure.
The human-rights purpose cl 6.1.3 is not aspiration — it binds the design. Clause 39 mandates privacy-by-design: minimised collection, minimised central custody, encryption and selective disclosure where feasible, and independent auditability. An identity system operated under this constitution cannot lawfully drift into a surveillance instrument, because the entity running it is constitutionally barred from building one.
Consent in VEID is scope-based and granular: global settings, per-scope consent with purpose, expiry, and revocation timestamps, and provider-specific grants with whitelists and restrictions. The published consent framework maps the implementation (x/veid/types/consent.go) against GDPR, CCPA/CPRA, and BIPA, and a biometric data addendum sets specific commitments for biometric handling. Biometric capture requires explicit consent before anything is collected.
The patent record is public: accepted Australian patent AU2024203136B2 covers the decentralized computing system including the identity-scoring blockchain control layer and on-chain multifactor authentication — held by the foundation to prevent private capture, while the implementation is published under Apache-2.0.
05Open questions
Identity is a research program, and its open problems are stated as such on the decentralized-identity topic page.
06Go deeper
