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

Back to watch register
8Privacy & surveillanceBelow the top 20

Workplace emotion recognition

AffordanceImpact domainBoth

Statement (NASA form)

Given that vendors already sell affect models to managers as ‘wellbeing’, there is a possibility of workers scored on faces and tone they cannot contest resulting in a new, unregulated personnel file that follows people between jobs.

Likelihood
3Probable
Consequence
2Minor
Urgency
2Planned

A feeling is not a performance metric. The product pretends it is.

Composite 8 = 3×2 + 2

Applicable mitigations

Related on the map