The Illusion of Reasoning: Exposing Evasive Data Contamination in LLMs via Zero-CoT Truncation

TL;DR

This paper introduces Zero-CoT Probe, a novel black-box method that detects evasive data contamination in large language models by truncating reasoning steps and comparing performance on original and perturbed datasets.

Contribution

It presents a new detection technique for stealthy data contamination in LLMs, including a metric for contamination severity and a publicly available implementation.

Findings

ZCP effectively detects both direct and evasive contamination.

It distinguishes memorization from problem-solving abilities.

The method is validated on various contaminated models.

Abstract

Large language models (LLMs) have demonstrated impressive reasoning abilities across a wide range of tasks, but data contamination undermines the objective evaluation of these capabilities. This problem is further exacerbated by malicious model publishers who use evasive, or indirect, contamination strategies, such as paraphrasing benchmark data to evade existing detection methods and artificially boost leaderboard performance. Current approaches struggle to reliably detect such stealthy contamination. In this work, we uncover a critical phenomenon: a model's generated reasoning steps actively mask its underlying memorization. Inspired by this, we propose the Zero-CoT Probe (ZCP), a novel black-box detection method that deliberately truncates the entire Chain-of-Thought (CoT) process to expose latent shortcut mappings. To further isolate memorization from the model's intrinsic problem-solving capabilities, ZCP compares the model's zero-CoT performance on the original benchmark against an isomorphically perturbed reference dataset. Furthermore, we introduce Contamination Confidence, a metric that quantifies both the likelihood and severity of contamination, moving beyond simple binary classifications. Extensive experiments on both previously identified contaminated models and specially fine-tuned contaminated models demonstrate that ZCP robustly detects both direct and evasive data contamination. The code for ZCP is accessible at https://github.com/Yifan-Lan/zero-cot-probe.