# Is Evaluation Awareness Just Format Sensitivity? Limitations of Probe-Based Evidence under Controlled Prompt Structure

arXiv: https://arxiv.org/abs/2603.19426 · HTML: https://mlsanity.com/paper/2603.19426/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Probing deception, bias and beliefs, one of 13 directions in Mechanistic interpretability.
- Builds on 8 of its 9 references that are in the map.
- First posted on arXiv 19 Mar 2026.

## Placed in

- Mechanistic interpretability / [Probing deception, bias and beliefs](https://mlsanity.com/mechinterp/probing-deception-bias-and-beliefs/)

## Builds on (most cited first, 8 of 8)

- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Measuring Massive Multitask Language Understanding](https://mlsanity.com/paper/2009.03300/) (2020)
- [Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge](https://mlsanity.com/paper/1803.05457/) (2018)
- [Large Language Models Often Know When They Are Being Evaluated](https://mlsanity.com/paper/2505.23836/) (2025)
- [MuTual: A Dataset for Multi-Turn Dialogue Reasoning](https://mlsanity.com/paper/2004.04494/) (2020)
- [Probing and Steering Evaluation Awareness of Language Models](https://mlsanity.com/paper/2507.01786/) (2025)
- [Steering Evaluation-Aware Language Models to Act Like They Are Deployed](https://mlsanity.com/paper/2510.20487/) (2025)
- [Evaluation Awareness Scales Predictably in Open-Weights Large Language Models](https://mlsanity.com/paper/2509.13333/) (2025)

## Built on by (most cited first, 3 of 3)

- [Evaluation Awareness Is Not One Capability: Evidence from Open Language Models](https://mlsanity.com/paper/2606.23583/) (2026)
- [Evaluation Awareness in Language Models: Representation, Verbalization, and Control](https://mlsanity.com/paper/2608.21766/) (2026)
- [Unread or Unenforced? Separating Representation from Enforcement Failure in Content Guards](https://mlsanity.com/paper/2609.26178/) (2026)

## Abstract (arXiv)

Prior work uses linear probes on benchmark prompts as evidence of evaluation awareness in large language models. Because evaluation context is typically entangled with benchmark format and genre, it is unclear whether probe-based signals reflect context or surface structure. We test whether these signals persist under partial control of prompt format using a controlled 2x2 dataset and diagnostic rewrites. We find that probes primarily track benchmark-canonical structure and fail to generalize to free-form prompts independent of linguistic style. Thus, standard probe-based methodologies do not reliably disentangle evaluation context from structural artifacts, limiting the evidential strength of existing results.