Introduction to:
Our primary testing services...
- Accuracy: Delivering precise and reliable results.
- Transparency: Our processes are open and clear, with detailed reporting and unbiased evaluations.
- Privacy: Protecting personal information is paramount. BixeLab follows strict data protection standards.
- Innovation: We improve testing methods to enhance efficiency and accuracy.
Our commitments:
Features of our lab...
Running a lab ...
Other Services
- Software:
- Performance Analysis and Assurance Tool (Performix)
- Quality Analysis Tools (OpenBQ)
- Training
- Manual Identity Resolution (BxAssure)
- Training Courses
- AI Assurance
- Decentalized ID (VC's, Passkeys etc)
- Assurance and Advice
- General AI (LLM etc)
- Testing and Assurance
- NGO work in Pacific and UNHCR
- Building stroing foundational identity system
Managing Manual Identity Resolution
- Selecting personnel with the right qualities and innate ability, suitable to undertake operator training.
- How to eliminate very poor performers?
- How to design team given available performance (especially at small scale)?
- Ensuring personnel have the right skills, knowledge and tooling to perform
- How to ensure good coverage of potential document fraud issues in remote examination?
- How to train on the mechanics of face discrimination? How to ensure insight is embedded about the kinds of exploits bad actors may attempt?
- Reaching fair outcomes free from biases such as confirmation and cognitive bias.
- How to train to minimize inherent biases in operators? How to test that these are eliminated?
- Compliance with necessary standards and protocols
- Get the right people on the team.
- Give human resolvers tools to support machine resolution for better onboarding outcomes.
Emerging Issues
AI & Identity
-
Deepfakes
→ threat to trust -
New Mobile Capabilities
→ e.g. contactless fingerprint -
Agents
→ new fraud vectors -
Wallets and Passkeys
- → new security threats
-
AI
→ new testing requirements - Data interchange
- Open Source Systems
AI Governance
Australia’s AI Ethics Principles
- Reliability & Saftey
AI Governance
Considerations
Some Strategic Discussion Points
- What will emerging threat vectors look like
- Deep fakes
- How to ensure high performance
- Systems
- People
- Impact of Passports in Wallets and On-Device Biometrics
- Age Assurance
- Watchlists
- Ongoing Training
- Bonafide:
- Flow 1 is more strict, more sensitive, and fails more often.
- Flow 2 is simpler, more forgiving, and passes more often.
- Attacks:
- Both flows correctly reject all presentation attacks (1420 total).
- Flow 1 creates all non-response issues.
- In bad lighting and with glasses glare, Flow struggles the most.
- Safari performs best across both flows.
- Chrome and Android show the highest failure rates in Flow 2.
Example Testing Outcomes
Copy of Copy of Copy of BixeLab Introduction
By Ted Dunstone
