Evan Hubinger
ON THE RECORD · 8 verbatim quotes · every quote source-verified
“To be clear, as we say in our latest Risk Report, I think the risk from present models is low. What I am worried about is superintelligence arising from recursive self-improvement, as we have said is happening faster than we thought.”
Evan Hubinger · said date not stated · reported 2026-09-15
Hubinger posted on X responding to Jacob Coxon's departure, clarifying his views on AI risk.
“We really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade,”
Evan Hubinger · said date not stated · reported 2026-09-13
Hubinger posted on X echoing Coxon's post about the dangers of AI.
“happening faster than we thought.”
Evan Hubinger · said date not stated · reported 2026-09-09
Hubinger added that recursive self-improvement is happening faster than previously thought.
“Jacob is correct here—we really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade. I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to,”
Evan Hubinger · said 2026-09-08 · reported 2026-09-09
Hubinger responded to Coxon's post on X, agreeing that AI could kill all humans and saying Anthropic lacks a plan to solve alignment for superintelligence.
“earnestly believe AI could kill all humans!”
Evan Hubinger · said date not stated · reported 2026-09-09
Hubinger echoed Coxon's sentiment, saying his team earnestly believes AI could kill all humans.
“superintelligence arising from recursive self-improvement,”
Evan Hubinger · said date not stated · reported 2026-09-09
Hubinger said the fear compounds with superintelligence arising from recursive self-improvement.
“have a plan to solve alignment for superintelligence and are not clearly on track to.”
Evan Hubinger · said date not stated · reported 2026-09-09
Hubinger admitted Anthropic does not have a plan to solve alignment for superintelligence and is not clearly on track to do so.
“What I am worried about is superintelligence arising from recursive self-improvement, as we have said is happening faster than we thought,”
Evan Hubinger · said 2026-09-08 · reported 2026-09-09
Hubinger added in another post that risks from current AI models are low but that superintelligence from recursive self-improvement is his concern.