AI Boss Tools | Diagnostic 04 • Back to the Hub
MIT's AI Risk Repository catalogues more than 1,700 documented AI risks and sorts them into seven domains. This tool walks you through all seven for your own organisation: you rate the likelihood and the impact of each, and it plots your risk matrix with a rating and a next step for every domain.
A risk matrix separates two questions that usually get mixed together: how likely a risk is, and how much damage it would do. Few organisations have applied one to AI, partly because there was no agreed list of what the risks actually are.
MIT's AI Risk Repository provides that list. First published by Slattery and colleagues in 2024 and updated since, it reviewed 74 risk frameworks and produced a single taxonomy of seven domains and 24 subdomains, covering discrimination, privacy, misinformation, misuse, and system failure among others.
You rate each domain twice: likelihood (1 to 5) and impact (1 to 5). Likelihood times impact gives a score out of 25, rated Low, Moderate, High, or Critical. The one-pager gives you a board-ready summary of all seven.
A living database of more than 1,700 AI risks drawn from 74 frameworks, classified into seven domains and 24 subdomains. The domain structure of this tool follows it directly.
The peer-reviewed meta-review behind the repository, from MIT FutureTech and collaborators.
A score for every dimension, a median benchmark to compare against, a specific next step for each dimension, and a one-pager sized for a board pack. Your name and email open it, and there is no charge.