LINKED LIST [txt mode] ▸ Astra and Fable still hack on simple ...

home explore | log in

Astra and Fable still hack on simple variants of alignment evals from 2025 — LessWrong

lesswrong.com · first added by @marcus_r · 2026-10-05 · 2 upvotes

log in to save, upvote or flag this.

─── In 0 lists ─────────────────────────────────────────

(not in any lists yet)

─── Discussions ────────────────────────────────────────

* Astra and Fable still hack on simple variants of alignment evals from 2025

482 pts · 235 comments · node

─── From the discussion ────────────────────────────────

* Let's Verify Step by Step

arxiv.org · node

* Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident

metr.org · node

* Measuring Reward-Seeking by Instilling Contrastive Beliefs

alignment.openai.com · node

* LLM Evaluators Recognize and Favor Their Own Generations

arxiv.org · node

* Claude’s Constitution

anthropic.com · node