Examining the differential risk from high-level artificial intelligence and the question of control
A structured, evidence-backed AI risk taxonomy can directly inform how APS agencies categorise and assess AI-related risks in governance frameworks.
Key points
- MIT AI Risk Repository summarises a four-class framework covering misuse, accident, structural, and agential AI risks.
- Expert survey data identifies monopolistic race dynamics, alignment failures, and power-seeking as highest-impact risks.
- A useful taxonomy for APS risk registers, though the framework targets advanced/AGI-level AI rather than current deployments.
Implications for Australian agencies
- Consider APS governance and risk practitioners could consider mapping this four-class taxonomy against existing agency AI risk registers to identify gaps or unstated assumptions.
- Monitor Teams tracking frontier AI safety may want to monitor the full MIT AI Risk Repository as a curated reference for emerging risk frameworks.
Implications are AI-generated. Starting points, not advice — see methodology for how they're framed.
View original source
Copied.
"Examining the differential risk from high-level artificial intelligence and the question of control"
Source: MIT AI Risk Repository – Blog
Published: 1 December 2024
URL: https://airisk.mit.edu/blog/examining-the-differential-risk-from-high-level-artificial-intelligence-and-the-question-of-control
The MIT AI Risk Repository has published a summary of a peer-reviewed paper by Kilian, Ventura, and Bailey (2023) that presents a hierarchical framework classifying advanced AI risks across four categories: misuse, accident, structural, and agential. Drawing on expert survey data from public and private sector domain experts, the framework identifies monopolistic race dynamics, AI alignment failures, and power-seeking behaviour as the highest-impact risk areas. It also uses exploratory modelling to characterise future scenarios under varying social and technological conditions. While the framework is oriented toward advanced and AGI-level systems, its taxonomy has practical utility as a reference for agencies developing AI risk registers or governance structures.
Implications for Australian agencies:
- [Consider] APS governance and risk practitioners could consider mapping this four-class taxonomy against existing agency AI risk registers to identify gaps or unstated assumptions.
- [Monitor] Teams tracking frontier AI safety may want to monitor the full MIT AI Risk Repository as a curated reference for emerging risk frameworks.
Retrieved from SIMS, 18 July 2026.