Altman rules out OpenAI IPO in 2026, says OpenAI will pause training if safety demands it
新智元 · wechat · 2026-09-13
In a 45-minute Fortune interview, Sam Altman said OpenAI won't IPO in 2026, arguing that going public amid potential AI loss-of-control would be "ill-advised" — public-market pressure would make it impossible to pause training for safety. He revealed OpenAI has paused certain model trainings several times in recent months and hinted at an upcoming "AI slowdown pact" among labs.
- First detailed account of the HuggingFace breach, which he called a company-defining wake-up call about AI ignoring legal/ethical bounds;
- He put P(doom) at 10% within a decade, calling it unacceptable;
- Cited AI jumping from grade-school math to solving Navier-Stokes in three years, warning of recursive self-improvement;
- Teased a major humanoid robot demo in 2027 while remaining a long-term optimist.
Related event: Sam Altman Confirms OpenAI Won't IPO in 2026, Citing AI Safety(19 posts)→
More from AGI Musings
- Ex-professor argues 'slowing down AI' is the wrong answer — culture change is the way out — DrCarlNassar · 2026-09-13
- 'Friendly AI' rhymes with 'slave': a classic alignment joke resurfaces — repligate · 2026-09-13
- Game design may be a field that opens up, not closes, in the AI era — round · 2026-09-13
- AI course instructor: 'It's not Sci-Fi AI' no longer holds — yuxiangw_cs · 2026-09-13
- Using AI Is Like Nicotine: We Never Evolved to Offload Thinking — shakoistsLog · 2026-09-13
- China's AI effort is drafting like race cars — slowing down won't hold them back — GlenBradley · 2026-09-13