Safety researcher pushes back: controlling non-godlike AI has been Redwood's core agenda since 2023

trevposts · x · 2026-09-25

Responding to claims that x-risk researchers are "dangerous" and distract from interventions like tightly controlled AI systems, the author notes that controlling increasingly powerful but not vastly superhuman AI has been the primary research agenda of Redwood Research since 2023 — an institution that co-authored the METR report. He adds it's frustrating to be labeled "dangerous" for distracting from interventions these researchers have long championed.

Original post →

More from AGI Musings

AGI Musings channel →