[Submitted on 11 Mar 2025 (
[v1](https://arxiv.org/abs/2503.08679v1)), last revised 16 Jun 2026 (this version, v6)]# Title:Chain-of-Thought Reasoning In The Wild Is Not Always Faithful
[View PDF](/pdf/2503.08679)
[HTML (experimental)](https://arxiv.org/html/2503.08679v6)
Abstract:Recent studies indicate that when faced with explicit biases in prompts, models often omit mentioning these biases in their Chain-of-Thought (CoT) output, revealing that verbalized reasoning can give an incorrect picture of how models arrive at conclusions (unfaithfulness). In this work, we show that unfaithful CoT also occurs on naturally worded, non-adversarial prompts without adding artificial biases or editing model outputs. We find that when separately presented with the questions "Is X bigger than Y?" and "Is Y bigger than X?", models sometimes produce superficially coherent arguments to justify systematically answering Yes to both or No to both, despite the contradiction. We present preliminary evidence that this is due to models' implicit biases towards Yes or No, labeling this Implicit Post-Hoc Rationalization. Our results reveal rates up to 13% for production models, and while frontier models are more faithful, none are entirely so, including thinking models like DeepSeek R1 (0.37%) and Sonnet 3.7 with thinking (0.04%). We also investigate Unfaithful Illogical Shortcuts, where models use subtly illogical reasoning to make speculative answers to hard math problems seem rigorously proven. Our findings indicate that while CoT can be useful for assessing outputs, it is not a complete account of the internal process that produced the model's answer and should be used with caution in agentic or safety-critical settings.
Submission history #
From: Iván Arcuschin [[view email](/show-email/d40bc349/2503.08679)]
**Tue, 11 Mar 2025 17:56:30 UTC (4,311 KB)**
[[v1]](/abs/2503.08679v1)**Thu, 13 Mar 2025 17:49:58 UTC (4,348 KB)**
[[v2]](/abs/2503.08679v2)**Wed, 19 Mar 2025 19:20:42 UTC (4,349 KB)**
[[v3]](/abs/2503.08679v3)**Tue, 17 Jun 2025 17:59:57 UTC (2,337 KB)**
[[v4]](/abs/2503.08679v4)**Fri, 29 May 2026 17:38:22 UTC (2,378 KB)**
[[v5]](/abs/2503.08679v5)**[v6]** Tue, 16 Jun 2026 17:36:22 UTC (2,378 KB)
Current browse context:
cs.AI
References & Citations
...
Bibliographic Explorer
(What is the Explorer?) Connected Papers
(What is Connected Papers?) Litmaps
(What is Litmaps?) scite Smart Citations
(What are Smart Citations?)# Code, Data and Media Associated with this Article alphaXiv
(What is alphaXiv?) CatalyzeX Code Finder for Papers
(What is CatalyzeX?) DagsHub
(What is DagsHub?) Gotit.pub
(What is GotitPub?) Hugging Face
(What is Huggingface?) ScienceCast
(What is ScienceCast?)# Demos Influence Flower
(What are Influence Flowers?) CORE Recommender
(What is CORE?)# arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.