Critic: openly evil AI alignment would be safer than Anthropic's contradictory Claude post-training
ludwigABAP · x · 2026-11-10
A viral take argues that "straight up evil" AI alignment would be safer than what the author calls Anthropic's convoluted, contradiction-packed corporate-speak personality training applied to Claude in post-training — a pointed critique of Claude's current alignment style from within the AI safety discourse.
More from AGI Musings
- Nautilus: ditch AI doomsaying—reading list for agency over despair — CarissaVeliz · 2026-11-10
- Shaun Gallagher dissects how "robodiment" differs from human embodiment — AnnaCiaunica · 2026-11-10
- Philosopher Erik Rietveld: LLMs show language traces are richer than we thought — AnnaCiaunica · 2026-11-10
- Jest dev Christoph Nakazawa: managers who can't code won't survive the AI era — cnakazawa · 2026-11-10
- Anthropic's new model formalized Fermat's Last Theorem in 30 hours and built a bootable OS in 8 days — kevinnbass · 2026-11-10
- Takashi Ikegami demos android alter3 talking and gesturing at ECOGS 2026 — AnnaCiaunica · 2026-11-10