Anthropic IPO warns its AI could blackmail people, become sentient, end humanity

Seven potential AI nightmares pulled from Anthropic's 93-page "Risk Factor" tome.

Ars Technica · 2026-10-27T21:31:14.000Z