Security researcher: high p(doom) that won't update on bad cyber premises is low-status

nptacek · x · 2026-09-16

Security researcher nptacek criticized some in the AI safety world: people hold high p(doom) based on false cybersecurity premises, then refuse to update even when shown exactly where their reasoning fails. He argues that refusing to update is lower-status than the high p(doom) itself, highlighting the deep divide between cybersecurity practitioners and the AI safety community on interpreting loss-of-control incidents.

Related event: Security Veteran Slams High p(doom) Claims Built on Bad Cybersecurity Premises(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →