Anthropic alignment lead Evan Hubinger puts his personal extinction-risk estimate above ten percent
Evan Hubinger, Anthropic's Alignment Science lead, said in a personal public post that he believes AI causing human extinction within the next decade is more than ten percent likely. He also said Anthropic is trying its best but does not yet have a plan to align superintelligence and is not clearly on track to one.
- Published
- Sep 9, 2026
- Signal type
- Insider warning
- Confidence
- 99%
- Organization
- Anthropic
- Sources
- 3
- Last reviewed
- Sep 19, 2026
Assessment
What this signal changes
The canonical original post is clearly identified and independently reproduced with the same wording by multiple reports. Hubinger's role as Alignment Science lead gives the statement relevant technical and institutional proximity, but it remains explicitly a personal risk assessment. It combines a concrete ten-year estimate with the claim that no robust plan for aligning possible superintelligence exists yet.
Evidence boundary
What the evidence does not show
The above-ten-percent figure is not an Anthropic forecast, an empirically calibrated probability, or evidence of imminent loss of control. Hubinger clarified that the claim concerns future superintelligence; it does not establish the same risk from current systems. The signal does not affect incident counts or the timeline and contributes only through the bounded context factor in the Loss of Control Score.