Zuck vs Dario on alignment: liability incentives already baked in, says Carter recap

naval · x · 2026-09-16

Naval amplifies Nic Carter's recap of Zuckerberg systematically rebutting Dario Amodei's alignment talking points, adding that labs face no real liability when things go wrong — the key gap to fix for illegal actions by AI agents.

Carter's recap: people want models aligned with them (a jab at Anthropic's constitutional approach); liability incentives already exist; Meta delayed Muse for alignment quietly; Anthropic may be trying to kingmake METR; alignment needs no coordination between labs.

Related event: Zuckerberg pushes back on AI slowdown: alignment is capability(15 posts)→

Original post →

More from AGI Musings

AGI Musings channel →