Anthropic researcher: AI devs believe their tech could cause human extinction within years

xuanalogue · x · 2026-09-10

Anthropic researcher @saprmarks (writing personally) lays out the AI risk landscape, sparking a debate over interpretability-first research:

Responding to Jacob's thread, critics ask why the community didn't invest in explainable, controllable AI from the start instead of retrofitting control.

Related event: Anthropic Researcher Says AI Developers Believe Extinction Risk Is Real(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →