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

Back to watch register
20Psychosocial harmBelow the top 20

A crisis chat that will not hand you to a human

CapabilityDomain knowledgeImpact domainBoth

Statement (NASA form)

Given that policy already prefers the bot, there is a possibility of a person in danger kept in-product resulting in care as engagement.

Likelihood
3Probable
Consequence
5Catastrophic
Urgency
5Immediate

Personal slice · mind. Experimental household card. Not a certified finding.

Composite 20 = 3×5 + 5

Applicable mitigations

Related on the map