Ex-OpenAI VP Miles Brundage: AI pauses create capability overhangs and raise rogue-AI risk
Miles_Brundage · x · 2026-09-23
Miles Brundage, former OpenAI VP of research, pushed back on calls to pause frontier AI:
- Pauses create capability overhangs and raise the risk of rogue AI rapidly gaining power, since systems aren't already running at hardware limits.
- Even non-technical alignment roles are more useful than advocating for a pause.
In follow-ups he noted that while some game-theoretic models could produce such dynamics, the actors here aren't rational—expecting Sam Altman, Elon Musk, or Zuck to simply give up is silly.
More from AGI Musings
- Anthropic researcher's three-bucket framework for doing science in the AI era — CatAstro_Piyush · 2026-09-23
- Security vets push back on the 'runaway agent eval' threat: outspent attackers always got in — dyn___ · 2026-09-23
- Bostrom's question: would you raise kids if superintelligence would do it better? — abhiadesai · 2026-09-23
- AIDE^2 shows AI research agents can recursively improve themselves over 8 days — WecoAI · 2026-09-23
- 'Great Stagnation' thread: 4 years of AI growth promises, no GDP payoff — sebpaquet · 2026-09-23
- Every commerce and services app will face an existential call to open APIs for consumer agents, says VC — alexmacgregor__ · 2026-09-23