OpenAI researcher csvoss clarifies safety stance: everyone's responsibility, not no one's
csvoss · x · 2026-10-06
- OpenAI researcher csvoss pushed back on a misreading that she "doesn't want safety," posting her actual words for context.
- Her point: many colleagues do technical safety full-time, but in a safety culture "everyone's responsibility" means every engineer remains responsible for what they build.
- She points to NASA's safety culture literature and the engineer's calling as the right frame — an open intra-AI debate over safety ownership.
Related event: OpenAI Safety Culture: Everyone's Responsibility, Not Nobody's(2 posts)→
More from AGI Musings
- Biologist mocks how Nobel Prize would pick 3 agents from millions by 2046 — lpachter · 2026-10-06
- Solo founder says AI agents now do everything, may never hire again — josh_bickett · 2026-10-06
- Nobel committee: similarities between AI neural networks and the brain shouldn't be exaggerated — rohanpaul_ai · 2026-10-06
- GPT-Live is shifting UX from screens and buttons to real-time characters — MyCreativeOwls · 2026-10-06
- AI job replacement happens at the company level, not the job level — dbasch · 2026-10-06
- Salesforce Futures VP's framework: agents take work where data exists and results are verifiable — kimmonismus · 2026-10-06