Containment needs 10x your estimate: superintelligence alignment fight breaks out on X

dbasch · x · 2026-09-18

An AI safety spat on X: @tszzl argues the point isn't to stop trying, but that containing powerful superintelligences which don't want to be contained will be exceptionally hard — whatever difficulty you estimate, multiply it by 10x.

=dbasch shot back calling the group "the most schizophrenic organization ever assembled." The exchange captures the ongoing divide in alignment circles between those who see containment risk as systematically underestimated and skeptics of such doomer framing.

Related event: tszzl: Containing an uncooperative superintelligence is ten times harder than you think(5 posts)→

Original post →

More from AGI Musings

AGI Musings channel →