The Risks Associated with Artificial General Intelligence: A Systematic Review
Agencies building AI risk frameworks can reference this synthesis to benchmark their own risk categories against emerging AGI-focused literature.
Key points
- A 2023 systematic review identifies six AGI risk categories, from unsafe goals to existential risks.
- The MIT AI Risk Repository is cataloguing risk frameworks - a useful reference for APS risk taxonomy work.
- The reviewed literature relies heavily on philosophical discussion; limited peer-reviewed empirical risk modelling exists.
Summary
This MIT AI Risk Repository blog post spotlights a 2023 systematic review (McLean et al.) that synthesises 16 articles on AGI risks, identifying six categories: loss of human control, unsafe goals, unsafe development, poor ethics and values, inadequate management, and existential risks. The review also critiques the state of AGI risk literature, noting reliance on philosophical discussion, limited peer-reviewed content, unclear definitions, and no standard terminology. The MIT AI Risk Repository is building a curated collection of such frameworks, which may be useful for agencies seeking to ground their own risk classification work in published research.
Implications for Australian agencies
- Monitor Risk and governance teams may want to monitor the MIT AI Risk Repository as a curated reference base when developing or reviewing agency AI risk taxonomies.
- Consider Agencies scoping long-horizon AI risk frameworks could consider whether the six AGI risk categories map usefully onto their existing risk registers.
Implications are AI-generated. Starting points, not advice.
"The Risks Associated with Artificial General Intelligence: A Systematic Review" Source: MIT AI Risk Repository – Blog Published: 17 December 2024 URL: https://airisk.mit.edu/blog/the-risks-associated-with-artificial-general-intelligence-a-systematic-review This MIT AI Risk Repository blog post spotlights a 2023 systematic review (McLean et al.) that synthesises 16 articles on AGI risks, identifying six categories: loss of human control, unsafe goals, unsafe development, poor ethics and values, inadequate management, and existential risks. The review also critiques the state of AGI risk literature, noting reliance on philosophical discussion, limited peer-reviewed content, unclear definitions, and no standard terminology. The MIT AI Risk Repository is building a curated collection of such frameworks, which may be useful for agencies seeking to ground their own risk classification work in published research. Implications for Australian agencies: - [Monitor] Risk and governance teams may want to monitor the MIT AI Risk Repository as a curated reference base when developing or reviewing agency AI risk taxonomies. - [Consider] Agencies scoping long-horizon AI risk frameworks could consider whether the six AGI risk categories map usefully onto their existing risk registers. Retrieved from SIMS, 18 May 2026.