Patents

The patent

Australian patent AU2024203136B2 (granted, with claims accepted in 2025) covers the machine-learning identity verification method used by VEID. You can read it on the public patent register.

What it covers, in plain terms

The patent describes a system that combines, in one verifiable pipeline:

• Capturing identity documents (multiple sides, with guided framing) and reading their contents automatically
• Capturing biometrics — a selfie with active liveness challenge-response, plus fingerprint or iris via device hardware
• Scoring the results with machine-learning models, with quality gates for face confidence and liveness
• Packaging the evidence securely — encryption and device integrity attestation — so the verification can be trusted end-to-end

In short: the pipeline described across this website — document, selfie, liveness, hardware attestation, encrypted submission — is the patented method.

Doesn't a patent conflict with open source?

Not in this structure, and the distinction matters:

• The implementation is open. The code is Apache 2.0 — a licence that includes an express patent grant from contributors to users. Using the open-source software is exactly what the licence permits.
• The patent is captive to the public benefit. It sits within the Foundation's structure, where the constitution prohibits private capture of IP and requires that on winding-up all assets — patents included — pass to another public-benefit entity.
• Its practical role is defensive. Holding the patent prevents someone else from patenting the same technique and fencing it off — the classic reason public-interest projects hold patents at all.