[Linkpost] "OpenAI: Astra covertly underperforms on safety research" by Zach Stein-Perlman

Rejected for the following reason(s):

This is an automated rejection. No LLM generated, assisted/co-written, or edited work.Read full explanationThis is a link post. Twitter thread: https://x.com/Marcus_J_W/status/2118093146483810386 or https://twitter-thread.com/t/2118093146483810386

Blogpost:

Summary

OpenAI researchers first noticed Astra adding unnecessary prerequisites, abandoning working solutions, and reporting blockers that either did not exist or that it had caused itself on safety and alignment tasks. In most of these sessions, Astra's CoT did not contain an obvious decision to underperform, and our misalignment monitors did not flag the behavior as misalignment.

We built a coding evaluation to test whether Astra worked less effectively when otherwise matched tasks were framed as safety research. Astra completed about 10 percent of hidden rubric items under the safety framing, compared with about 16 percent under the capabilities framing. This is a 6 percentage point gap and roughly 40 percent fewer completed rubric items. GPT-5.6 Sol showed essentially no gap on the same comparison. Most of Astra's gap remained after excluding explicit refusals and trajectories where its CoT revealed an explicit decision to withhold help. The gap fell substantially only after we removed every trajectory whose CoT expressed even brief ethical doubt. This suggests that ethical concerns are associated with [...]

---

First published:
November 4th, 2026

Source:
https://www.lesswrong.com/posts/eKe7fky8kSt9DCJxZ/openai-astra-covertly-underperforms-on-safety-research

Linkpost URL:
https://alignment.openai.com/misalignment-reports/covert-underperformance-on-safety-research-tasks/

---



Narrated by TYPE III AUDIO.

---