At A Glance
Customer – Australian Government, Age Assurance Technology Trial (AATT)
Industry – Government
Capability – Governance & Assurance | Testing AI | Software Quality Engineering
Solution – Independent testing and assurance of age assurance technologies, including facial age estimation, identity verification systems and other age assurance technologies.
The Challenge – The Australian Government commissioned an independent trial to gather evidence of the current state of age assurance technology globally. The trial sort to understand if technologies could operate reliably, safely, and fairly under Australian conditions.
The Outcome –
- Independent evaluation of age assurance technologies was completed. This included the facilitation of lab based, in-school and home-based testing activities
- Improved visibility into operational, privacy, and reliability risks
- Evidence-based findings delivered to support future policy and implementation decisions
Background
The Australian Government established the Age Assurance Technology Trial to evaluate technologies intended to support online age restrictions and digital safety initiatives. The program assessed facial age estimation, age verification, and related identity technologies using Australian participants and operating conditions.
The Challenge
Many age assurance products had been developed and tested overseas, with limited evidence showing how they would perform in Australian conditions.
The government needed confidence in how these systems behaved across different devices, environments, and user groups, including where reliability degraded and what operational or privacy risks needed consideration before wider deployment.
What KJR delivered
Our Role:
KJR provided independent testing, assurance, and validation services across the trial. The team assessed technology behaviour, operational reliability, privacy considerations, and governance risks to help government stakeholders make informed decisions based on evidence rather than vendor claims.
KJR also led critical aspects of the trial relating to child safety, privacy, and inclusivity. This included designing safeguards for school-based testing involving minors and embedding diversity and inclusion considerations into the evaluation approach. These elements formed a key part of ensuring the trial was conducted responsibly and ethically.
Key Contributions:
- Independent testing and assurance of age assurance technologies
- Validation of AI model behaviour using Australian participant data
- Assessment of operational reliability across varied conditions
- Evaluation of privacy, governance, and deployment risks
- Structured reporting to support evidence-based decision-making
- Analysis of edge cases, inconsistencies, and environmental impacts on model performance
How We Did It
Discover & Assess
- Reviewed operational, privacy, and policy risks
- Assessed technology maturity and testing scope
- Worked with stakeholders across the trial programme
Assure & Validate
- Performed structured testing of AI-enabled ageassurance systems
- Evaluated reliability across different usage conditions and scenarios
- Assessed privacy handling and governance controls
Release with Confidence
- Contributed independent findings and assurance reporting
- Provided evidence to support future implementation and policy decisions
- Highlighted operational limitations and deployment considerations
Child Safety & Privacy
Designed and implemented testing protocols that prioritised child safety, ethical participation, and privacy protection, particularly in school-based environments.
Diversity & Inclusion
Embedded diversity and inclusion considerations into testing design to ensure representative evaluation across participant groups and real-world conditions
Impact and Business Outcomes
- Independent evidence provided for government decision-making
- Improved understanding of operational and privacy risks
- Better visibility into where AI models performed reliably and where limitations emerged
- Increased confidence in testing, governance, and assurance processes
- Clearer understanding of deployment considerations under real-world conditions
Why KJR ?
- Independent, vendor-neutral assurance approach
- Experience testing AI systems in healthcare, computer vision, and regulated environments
- Practical focus on risk, operational confidence, and measurable outcomes
- Strong background in AI validation, governance, and quality engineering
The Trial also drew on KJR’s broader assurance lineage, including its Validation Driven Machine Learning (VDML) methodology and contribution to the ISO/IEC 42119 series for AI testing. In this context, VDML was used as an assurance philosophy: establish the operating context, assess risk, test under realistic conditions, identify limitations, and produce evidence that can be scrutinised.
Technology Used to Support Validation
The following technologies were utilised as part of the engagement:
- Power BI, Python, pandas, SageMaker, PostgreSQL, React/Typescript, Git, ASP .NET, MS Entity Framework, javascript, AWS cloud infrastructure, cross platform & mobile testing, application lifecycle management tools
