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

Back to watch register
15Cyber & softwareBelow the top 20

Population-scale spearphish

CapabilityDomain knowledgeAffordanceImpact domainCap-adjacent

Statement (NASA form)

Given that models can write a personalised lure from a public profile in seconds, there is a possibility of millions of distinct, high-quality phishing messages in a day resulting in credential harvest that no longer looks like a campaign.

Likelihood
4Likely
Consequence
3Major
Urgency
3Priority

Spam filters were built for copies. These are not copies.

Composite 15 = 4×3 + 3

Applicable mitigations

Related on the map