OpenAI Chief Scientist: No Lab's Alignment Ready for Full-Speed Scaling, Calls for Voluntary Slowdown
OpenAI Chief Scientist Jakub Pachocki publicly stated that no lab today has solved alignment and oversight well enough to responsibly scale at full speed over a longer horizon, and he expects and hopes that "voluntary slowdown" becomes the norm until the industry establishes shared safety standards; he also said that once you truly grasp the stakes, the idea of "racing ahead at all costs" seems absurd. The remarks coincide with researcher merettm's long-form piece "An Alien Mind," sparking wide discussion and worth attention.
Confirmed
- OpenAI's website published the long-form essay "An Alien Mind" by merettm, surveying the current state of AI, concerns about AI development in the coming years, and the choices humanity must make to keep the future in its own hands; OpenAI also released another article the same day about LLMs accelerating science.
- Pachocki stated explicitly: no lab has alignment and oversight solved to the point of responsibly continuing to scale at maximum speed; he expects and hopes voluntary slowdown becomes the norm until shared safety standards are established; he also urged governments to pay attention to AI development.
- Pachocki called the idea of "sprinting ahead at all costs" absurd—remarks amplified by Gary Marcus and others, fueling debate over AI safety.
- On RSI, Pachocki strongly expects, based on internal results, that the current pace of progress can carry into "recursive self-improvement" (RSI); OpenAI has organized research around this goal, viewing RSI as key to staying at the frontier; meanwhile, he noted that the reliability of chain-of-thought monitoring is declining.
- Multiple community members joined the discussion: Nathan Calvin (Encode advocate) worried that many in the industry haven't yet grasped the risks; ericho said "understanding this alien mind is the most important problem in the world"; researcher K. Liu pointed out that relying on voluntary deceleration while maintaining rapid scaling won't last long.
Why it matters
- Coming from one of OpenAI's top research leaders, this call for the whole industry to slow scaling amid an intense frontier race stands in tension with the company's own commercial pace—a rare public expression of the safety camp's internal stance.
- The coexistence of RSI expectations with a judgment that alignment falls short means OpenAI may be betting on recursive self-improvement while acknowledging that existing oversight methods (such as chain-of-thought monitoring) are becoming less reliable—directly bearing on the risk boundary of frontier model development.
- The Reddit community has connected this to discussions of inter-lab scaling slowdowns and international coordination, and amplification by critics like Gary Marcus has spread the debate to a broader public.
2026-09-07 ~ 2026-09-07 · 11 related posts
Primary sources
- An Alien Mind: essay warns AI's next few years could slip out of humanity's hands — morqon · 2026-09-07
- Alignment researcher: no lab has solved alignment well enough to keep max-speed scaling responsibly — willdepue · 2026-09-07
- [source] OpenAI chief scientist: no lab has solved alignment enough to keep max-speed scaling — GaryMarcus · 2026-09-07
- OpenAI's chief scientist calls racing ahead at all costs 'absurd' as safety concerns mount — GaryMarcus · 2026-09-07
- OpenAI chief scientist Jakub Pachocki expects progress toward recursive self-improvement as CoT monitoring reliability declines — Hesamation · 2026-09-07
- [source] OpenAI Chief Scientist: No Lab Has Solved Alignment Enough to Keep Scaling at Max Speed — Tinac4 · 2026-09-07
- [source] OpenAI Publishes 'An Alien Mind' by Its Chief Scientist — bparrish · 2026-09-07
- OpenAI Chief Scientist Publishes 'An Alien Mind' on RSI Risks and Slowing Scaling — Bloated_Plaid · 2026-09-07
- OpenAI chief scientist: racing ahead at all costs is absurd once you grasp the stakes — AdrienLE · 2026-09-07