# OpenAI breakthrough revives fears that AI could become uncontrollable

https://mins.news/en/item/52102

- CryptoSlate, 2026-09-09 22:40 UTC
- Science
- Original: https://cryptoslate.com/openai-just-showed-why-one-of-its-former-researchers-thinks-ai-could-kill-everyone/

> OpenAI says an internal system coordinated about 10,000 AI agents to solve the Navier-Stokes existence and smoothness problem in 88 hours, with GPT-6 Astra later formalizing and verifying the result in Lean. The proof still requires wider scrutiny, but the claimed advance has intensified concerns about the widening gap between private frontier models, public tools and the risks of runaway AI.

## Related coverage

- [OpenAI claims Navier–Stokes breakthrough as calls for AI regulation grow](https://mins.news/en/video/52122) — Global News, 2026-09-09
- [OpenAI says 10,000 AI agents solved a $1 million math problem. Mathematicians are pushing back](https://mins.news/en/item/51217) — CoinDesk, 2026-09-09
- [Drama swirls around OpenAI’s claimed Navier-Stokes breakthrough](https://mins.news/en/item/50372) — The Verge, 2026-09-08