So8res: Anthropic fails its ideals as Dario dodges extinction-risk questions
trevposts · x · 2026-11-23
On the eve of Anthropic's IPO, AI safety researcher So8res publishes a long critique arguing the company does more to downplay AI dangers than to slow the race. He dissects Dario Amodei's media interviews: asked by Anderson Cooper whether AI could kill all humans, Dario deflected to Jacob Coxon's resignation and called the odds "not all that high"; asked by Jo Ling Kent about extinction risk, he pivoted to benefits. So8res contrasts these with direct answers Dario could have given, concluding he was "dissembling, not lying"—telling a newly attentive public "it's fine, go back to sleep."
More from AGI Musings
- If AI Does the Work, Who Gets to Learn? Lessons from Netscape to Anthropic's $81.3B IPO — shivsingh · 2026-11-23
- The great convergence: AI is turning every role and tool into the same builder — bendee983 · 2026-11-23
- WSJ: More twentysomethings skip entry-level jobs to become their own bosses — himanshustwts · 2026-11-23
- Economist: AI ends the premium on technical grunt work — asking questions matters more now — _akpiper · 2026-11-22
- Experts' "AI Will Never Do X" Predictions Keep Getting Falsified — Glen_Garrett_Gayhart · 2026-11-22
- WaPo investigation: Amodei privately put 20-30% odds on AGI destroying humanity — GaryMarcus · 2026-11-22