Debate: Is Anthropic's alignment team critically understaffed despite safety talk?
kuza55 · x · 2026-09-09
A debate over Anthropic's internal safety organization: one commenter alleges that despite Dario's safety rhetoric, the alignment team is critically understaffed and resourced inconsistently with stated concerns, arguing he should say so publicly. Quoted Joshua Saxe disagrees, noting internal safety teams can exist to make a company look good rather than to make the product safe — calling the issue complicated. The exchange touches on a core industry question about the independence of corporate safety teams.
More from Companies & People
- Oxford Codex meetup pairs 30 years of Bayesian ML with coding agents, $50 credits — paw_lean · 2026-09-09
- Ken Stanley on curiosity and not forcing outcomes, in conversation with VC Scott Sage — kenneth0stanley · 2026-09-09
- Tailwind Labs Is Joining Shopify, With 110M+ Weekly Installs — jh3yy · 2026-09-09
- YC S26 Demo Day Preview: Floating Data Centers, Diamond Chips, and 20,000 Tokens/sec Hardwired AI — bosmeny · 2026-09-09
- China's 2nd GEO Conference heads to Shenzhen Nov 21-22 with 399-yuan tickets, whitepaper debut — vista8 · 2026-09-09
- Thomson Reuters' 397B Qwen-based legal LLM beats GPT-5.4 on domain factuality — DeepLearningAI · 2026-09-09