Skip to main content
TPDI logo
Back to Layer 5

Capability 5.3.1

Assurance & Risk Management (safety, bias, explainability)

Definition

Skills to test, monitor and certify AI systems throughout their lifecycle, ensuring they are safe, fair, transparent, and compliant with laws and standards. This includes risk management, bias detection, safety testing and explainability audits.

Examples

  • National Framework for the Assurance of Artificial Intelligence in Government (2024)
  • International Association of Privacy Professionals (IAPP) Artificial Intelligence Governance Professional (AIGP) certification (delivered in Australia through Salinger Privacy, endorsed by DISR)
  • CSIRO Data61 National AI Centre Responsible AI Network (RAIN) evaluation and testing collaborations.

Indicators

  • Number of certified AI assurance or governance professionals (AIGP, ISO/IEC 42001 auditors)
  • Percentage of high-risk AI systems undergoing independent evaluation before deployment
  • Average time to resolve safety/bias issues found in audits

National AI Plan

PLAN COMMITMENT

Moderate commitment

Supports safety, bias, explainability through AI safety research, establishment of the AISI and guidance on responsible AI adoption, transparency and governance (Action 7, Action 8). Investment is modest compared to other jurisdictions, and the plan includes nothing specific on skills in this domain. Moderate commitment by government in the plan aligns roughly with the recommendation to close this critical gap, although greater attention is recommended.

Stocktake

Is the Stocktake missing evidence?

Share additional information for potential future Assessments