Ex-DeepMind, now Anthropic researcher: no viable scientific plan for recursively self-improving AI risks

harris_edouard · x · 2026-09-10

A researcher who worked at Google DeepMind and now at Anthropic (writing in personal capacity) says a common sentiment among peers is that there is not yet a viable scientific plan to solve risks from recursively self-improving AI.

Paul Graham shared it, arguing this is why American labs leading is a good thing—employees at Chinese labs couldn't warn about such threats, though Chinese labs will face the same threats within a year, in complete silence.

Original post →

More from AGI Musings

AGI Musings channel →