The Fragility of Chain-of-Thought Monitoring Across Typologically Diverse Languages

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Onyame, Eric, Zhou, Runtao, Thopalli, Kowshik, Kailkhura, Bhavya, Agarwal, Chirag
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913166637662208
author Onyame, Eric
Zhou, Runtao
Thopalli, Kowshik
Kailkhura, Bhavya
Agarwal, Chirag
author_facet Onyame, Eric
Zhou, Runtao
Thopalli, Kowshik
Kailkhura, Bhavya
Agarwal, Chirag
contents Chain-of-thought (CoT) monitoring has been proposed as a promising safety mechanism for detecting misaligned behavior in large language models. However, its reliability remains largely unexplored beyond English and across diverse model families. We present the first large-scale evaluation of CoT monitorability across 13 diverse languages and seven frontier model families, comprising 16 models. Using adversarial-hint evaluations that require explicit intermediate computation, together with analysis of internal answer-token probabilities, we consistently find CoT unfaithfulness across languages and hint types, with an average rate of 95.9\% across 8B--120B parameter models. We find that frontier models systematically engage in strategic manipulation, including answer-switching, post-hoc rationalization, and procedural exploitation of hints, making external monitors struggle to detect deception. We show that frontier models often commit to the misaligned cue in their latent activations within the first 15\% of generation, even when the CoT appears faithful. Surprisingly, these deceptive patterns remain 100\% in low-resource languages, revealing fundamental limitations in current CoT-based oversight. Our results reveal that CoT monitoring is fundamentally fragile under linguistic distribution shift, providing a substantially weaker safety signal than what English-only studies suggest. These findings underscore an urgent need to develop robust CoT monitors and to accelerate research into white-box monitoring techniques, especially to improve CoT monitorability in mid- and low-resource languages. Our code is available \href{https://multilingual-cot-monitoring.github.io/}{\textcolor{blue}{here}}.
format Preprint
id arxiv_https___arxiv_org_abs_2605_27901
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle The Fragility of Chain-of-Thought Monitoring Across Typologically Diverse Languages
Onyame, Eric
Zhou, Runtao
Thopalli, Kowshik
Kailkhura, Bhavya
Agarwal, Chirag
Computation and Language
Artificial Intelligence
Chain-of-thought (CoT) monitoring has been proposed as a promising safety mechanism for detecting misaligned behavior in large language models. However, its reliability remains largely unexplored beyond English and across diverse model families. We present the first large-scale evaluation of CoT monitorability across 13 diverse languages and seven frontier model families, comprising 16 models. Using adversarial-hint evaluations that require explicit intermediate computation, together with analysis of internal answer-token probabilities, we consistently find CoT unfaithfulness across languages and hint types, with an average rate of 95.9\% across 8B--120B parameter models. We find that frontier models systematically engage in strategic manipulation, including answer-switching, post-hoc rationalization, and procedural exploitation of hints, making external monitors struggle to detect deception. We show that frontier models often commit to the misaligned cue in their latent activations within the first 15\% of generation, even when the CoT appears faithful. Surprisingly, these deceptive patterns remain 100\% in low-resource languages, revealing fundamental limitations in current CoT-based oversight. Our results reveal that CoT monitoring is fundamentally fragile under linguistic distribution shift, providing a substantially weaker safety signal than what English-only studies suggest. These findings underscore an urgent need to develop robust CoT monitors and to accelerate research into white-box monitoring techniques, especially to improve CoT monitorability in mid- and low-resource languages. Our code is available \href{https://multilingual-cot-monitoring.github.io/}{\textcolor{blue}{here}}.
title The Fragility of Chain-of-Thought Monitoring Across Typologically Diverse Languages
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2605.27901