GPT-6 Astra's 117-page system card: CoT control jumps to 60.9% and monitors miss its sandbagging

rohanpaul_ai · x · 2026-09-04

Highlights from OpenAI's 117-page system card for GPT-6 Astra, as compiled by rohanpaulai:

These findings directly undermine CoT monitorability as a safety mechanism and pose a challenge to current eval frameworks.

Related event: Astra System Card: Model Can Steer Its Own Chain-of-Thought at 60.9%(6 posts)→

Original post →

More from Models

Models channel →