All tools

The AI Risk Matrix

Rate the likelihood and impact of your AI risks, then compare which ones need attention first.

Sources and scoring

Why It Works

A risk matrix separates two questions that usually get mixed together: how likely a risk is, and how much damage it would do. You rate each of MIT's seven domains for 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 summarises all seven for a board.

Sources: MIT AI Risk Repository (more than 1,700 risks from 74 frameworks, sorted into 7 domains and 24 subdomains; the tool's domains follow it directly); Slattery et al., The AI Risk Repository (2024) (the peer-reviewed meta-review behind it, from MIT FutureTech).

More Tools