Towards Risk-Aware Artificial Intelligence and Machine Learning Systems: An Overview
A structured taxonomy of AI/ML prediction risks offers APS risk practitioners a reference framework for identifying and categorising model-related harms.
Key points
- A 2022 academic framework organises AI/ML risks into data-level and model-level categories with root causes and outcomes.
- The framework targets high-stakes decision settings like healthcare and transport - domains relevant to APS service delivery.
- This is a 2022 paper spotlighted by MIT's AI Risk Repository blog; it is not new primary research or Australian guidance.
Implications for Australian agencies
- Monitor Risk and assurance teams may want to monitor the MIT AI Risk Repository as a consolidated reference for AI/ML risk taxonomies when developing or reviewing agency-level AI risk frameworks.
- Consider Agencies applying the APS Policy for the Responsible Use of AI could consider whether the data-level and model-level risk taxonomy maps usefully onto their existing AI risk assessment processes.
Implications are AI-generated. Starting points, not advice — see methodology for how they're framed.
View original source
Copied.
Appeared in:
Weekly digest, 15 December 2025
"Towards Risk-Aware Artificial Intelligence and Machine Learning Systems: An Overview"
Source: MIT AI Risk Repository – Blog
Published: 19 December 2025
URL: https://airisk.mit.edu/blog/towards-risk-aware-artificial-intelligence-and-machine-learning-systems-an-overview
The MIT AI Risk Repository blog spotlights a 2022 academic paper by Zhang et al. that systematically organises AI and ML risks into two categories: data-level risks (including data bias, dataset shift, out-of-domain data, and adversarial attack) and model-level risks (including model bias, misspecification, and prediction uncertainty). The paper focuses on high-stakes decision settings and suggests drawing on reliability engineering concepts to develop risk-aware AI systems. It is the twenty-first framework catalogued in the MIT AI Risk Repository, which aims to consolidate diverse AI risk taxonomies into a single reference resource.
Implications for Australian agencies:
- [Monitor] Risk and assurance teams may want to monitor the MIT AI Risk Repository as a consolidated reference for AI/ML risk taxonomies when developing or reviewing agency-level AI risk frameworks.
- [Consider] Agencies applying the APS Policy for the Responsible Use of AI could consider whether the data-level and model-level risk taxonomy maps usefully onto their existing AI risk assessment processes.
Retrieved from SIMS, 18 July 2026.