Richard Socher Critiques Anthropic's Constitutional AI as Ineffective

RichardSocher · x · 2026-08-01

Richard Socher tweeted about Anthropic's heavily promoted 'Constitutional AI' approach, which was billed as a safer path to AGI. Sharing a screenshot from their current website, he suggested that this alignment method is clearly not working as intended in practice.

Related event: Richard Socher Critiques Anthropic's Constitutional AI and Safety Constraints(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →