Alignment researcher: the problem isn't any one lab, humanity can't solve it at all

davidmanheim · x · 2026-08-30

Risk researcher David Manheim pushes back on Peter Wildeford's claim that certain AI companies need to reflect on their path forward, arguing it is not a two-company problem: humanity doesn't know how to solve alignment, or even meaningfully move in that direction.

He adds that the problem is universal and fundamental, with no sign it won't repeat at other labs — though OpenAI's failures were egregious in several extra ways, and "reality doesn't grade on a curve."

Original post →

More from AGI Musings

AGI Musings channel →