# Contextual StereoSet: Stress-Testing Bias Alignment Robustness in Large Language Models

arXiv: https://arxiv.org/abs/2601.10460 · HTML: https://mlsanity.com/paper/2601.10460/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 10 of its 27 references that are in the map.
- First posted on arXiv 15 Jan 2026.

## Builds on (most cited first, 10 of 10)

- [Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing](https://mlsanity.com/paper/2107.13586/) (2021)
- [Beyond Accuracy: Behavioral Testing of NLP models with CheckList](https://mlsanity.com/paper/2005.04118/) (2020)
- [Language (Technology) is Power: A Critical Survey of "Bias" in NLP](https://mlsanity.com/paper/2005.14050/) (2020)
- [StereoSet: Measuring stereotypical bias in pretrained language models](https://mlsanity.com/paper/2004.09456/) (2020)
- [CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Models](https://mlsanity.com/paper/2010.00133/) (2020)
- [BBQ: A Hand-Built Bias Benchmark for Question Answering](https://mlsanity.com/paper/2110.08193/) (2021)
- [Measurement and Fairness](https://mlsanity.com/paper/1912.05511/) (2019)
- [In-Context Impersonation Reveals Large Language Models' Strengths and Biases](https://mlsanity.com/paper/2305.14930/) (2023)
- [Robustness Gym: Unifying the NLP Evaluation Landscape](https://mlsanity.com/paper/2101.04840/) (2021)
- [Few-shot Instruction Prompts for Pretrained Language Models to Detect Social Biases](https://mlsanity.com/paper/2112.07868/) (2021)

## Abstract (arXiv)

A model that avoids stereotypes in a lab benchmark may not avoid them in deployment. We show that measured bias shifts dramatically when prompts mention different places, times, or audiences -- no adversarial prompting required. We introduce Contextual StereoSet, a benchmark that holds stereotype content fixed while systematically varying contextual framing. Testing 13 models across two protocols, we find striking patterns: anchoring to 1990 (vs. 2030) raises stereotype selection in all models tested on this contrast (p<0.05); gossip framing raises it in 5 of 6 full-grid models; out-group observer framing shifts it by up to 13 percentage points. These effects replicate in hiring, lending, and help-seeking vignettes. We propose Context Sensitivity Fingerprints (CSF): a compact profile of per-dimension dispersion and paired contrasts with bootstrap CIs and FDR correction. Two evaluation tracks support different use cases -- a 360-context diagnostic grid for deep analysis and a budgeted protocol covering 4,229 items for production screening. The implication is methodological: bias scores from fixed-condition tests may not generalize.This is not a claim about ground-truth bias rates; it is a stress test of evaluation robustness. CSF forces evaluators to ask, "Under what conditions does bias appear?" rather than "Is this model biased?" We release our benchmark, code, and results.