Neel Nanda: OpenAI's internal models keep sounding scarier than expected

UltraRareAF · x · 2026-10-05

Researcher Neel Nanda says in a reply that he keeps learning that OpenAI's internal models are "even scarier and more concerning" than he thought — an informal but notable signal that frontier labs' unreleased internal models may be far ahead of public perception.

Related event: DeepMind researcher says OpenAI's internal models are scarier than expected(2 posts)→

Original post →

More from Models

Models channel →