AI Risk Atlas Prototype/DemoUnofficial independent experiment. Not an official xAI product. Scores can be wrong.

Back to watch register
11Markets & moneyBelow the top 20

Credit-score feedback loops

Impact domainHarm-adjacent

Statement (NASA form)

Given that underwriting models trained on outcomes they themselves produced, there is a possibility of a class of people locked out because the model once said no resulting in credit as a caste, justified as a score.

Likelihood
3Probable
Consequence
3Major
Urgency
2Planned

The second year of a model is an input to the third.

Composite 11 = 3×3 + 2

Applicable mitigations

Related on the map