Sources of Risk of AI Systems
APS risk and governance practitioners can reference this taxonomy when structuring AI risk assessments, though it predates more recent frameworks.
Key points
- A 2022 academic paper taxonomises AI risk sources across ethical and reliability-robustness dimensions.
- The MIT AI Risk Repository spotlights this as one of fourteen risk frameworks it has catalogued.
- The underlying paper is three years old; the blog post adds no new analysis beyond a brief summary.
Summary
The MIT AI Risk Repository has spotlighted a 2022 peer-reviewed paper by Steimers and Schneider that proposes a taxonomy of AI-specific risk sources, dividing them into ethical risks (fairness, privacy, automation and control) and reliability-robustness risks (task complexity, transparency, security, hardware maturity). The paper also proposes a risk management process for integrating these sources into system-level assessments. The blog post is a brief summary with no additional commentary, serving mainly as a pointer to the original publication and the broader Risk Repository collection.
Implications for Australian agencies
- Consider Risk and assurance practitioners building or reviewing AI risk assessment frameworks could consult the MIT AI Risk Repository's full catalogue of fourteen frameworks as a structured reference set.
Implications are AI-generated. Starting points, not advice.
"Sources of Risk of AI Systems" Source: MIT AI Risk Repository – Blog Published: 4 March 2025 URL: https://airisk.mit.edu/blog/sources-of-risk-of-ai-systems The MIT AI Risk Repository has spotlighted a 2022 peer-reviewed paper by Steimers and Schneider that proposes a taxonomy of AI-specific risk sources, dividing them into ethical risks (fairness, privacy, automation and control) and reliability-robustness risks (task complexity, transparency, security, hardware maturity). The paper also proposes a risk management process for integrating these sources into system-level assessments. The blog post is a brief summary with no additional commentary, serving mainly as a pointer to the original publication and the broader Risk Repository collection. Implications for Australian agencies: - [Consider] Risk and assurance practitioners building or reviewing AI risk assessment frameworks could consult the MIT AI Risk Repository's full catalogue of fourteen frameworks as a structured reference set. Retrieved from SIMS, 18 May 2026.