Report: OpenAI uses opaque loop transformers, sparking safety concerns

sjgadler · x · 2026-09-02

A report suggests OpenAI and others are quietly adopting "loop transformers" that do not reveal their internal reasoning or "thinking" when scaled up. While this represents a leap forward in performance, it has sparked concerns regarding security both inside and outside of OpenAI, as increased power coupled with reduced opacity raises significant risks.

Related event: OpenAI's Astra Reportedly Uses Recurrent Depth Architecture, Hiding Its Reasoning and Raising Safety Alarms(35 posts)→

Original post →

More from Models

Models channel →