Researchers from OpenAI and Anthropic publish papers defining quantitative metrics for faithfulness and monitorability in chain-of-thought reasoning by large language models

— Related OpenAI trace spans 125 pages with one insight labeled frightening.

Original post

OPWD#254will depue|@WILLDEPUE

@littmath @alkalinesec There’s some great research from OpenAI and Anthropic here! https://arxiv.org/abs/2512.18311 https://arxiv.org/abs/2307.13702

2:05 PM · May 20, 2026 View on X