Anthropic's Evan Hubinger: AI extinction risk >10% this decade, alignment unsolved

SatelliteNetSec · x · 2026-09-10

Responding to a debate over data-center construction, Anthropic researcher Evan Hubinger said the company sincerely believes AI could kill all humans — his personal estimate is >10% within the next decade. He says Anthropic is trying its best, but there is no plan yet to solve alignment for superintelligence and the field isn't clearly on track. The quote drew both sarcasm (the best case is mass job loss and oligarchy) and calls from open-source advocates for 100x more research through open models and training data.

Related event: Anthropic Researcher's High-Profile Exit Over AI Doom Sparks Conspiracy Claims and Safety Debate(238 posts)→

Original post →

More from AGI Musings

AGI Musings channel →