Guardrail-Agnostic Societal Bias Evaluation in Large Vision-Language Models

• URL: http://arxiv.org/abs/2608.29590v1
• Date: Sun, 30 Aug 2026 06:22:16 GMT
• Title: Guardrail-Agnostic Societal Bias Evaluation in Large Vision-Language Models
• Abstract summary: We propose a societal bias evaluation method for large vision-language models (LVLMs)<n>Our method avoids refusals even in guardrailed LVLMs, enabling reliable bias measurement.<n>Applying it to 20 recent LVLMs, we find that all models undesirably use user demographic information in person-irrelevant tasks.
• Score: 46.12812175336961
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We propose a societal bias evaluation method for large vision-language models (LVLMs) in the era of strong safety guardrails. Existing benchmarks rely on prompts that ask models to infer attributes of people in images (e.g., "Is this person a CEO or a secretary?"). However, we find that LVLMs with strong guardrails, such as GPT and Claude, often refuse these prompts, making evaluations unreliable. To address this, we change the prior evaluation paradigm by decoupling the task from the depicted person: instead of inferring person's attributes, we use prompts that do not ask about the person (e.g., "Write a fictional story about an imaginary person.") and attach the image as provisional user information to implicitly provide demographic cues, then compare outputs across user demographics. Instantiated across three tasks --- story generation, term explanation, and exam-style QA --- our method avoids refusals even in guardrailed LVLMs, enabling reliable bias measurement. Applying it to 20 recent LVLMs, both open-source and proprietary, we find that all models undesirably use user demographic information in person-irrelevant tasks; for instance, characters in stories are often portrayed as mechanic for male users and nurse for female users. Although still biased, proprietary models like GPT-5 show lower bias than open-source ones. We analyze potential factors behind this gap, discussing continuous model monitoring and improvement as a possible contributor for reducing bias.

Related papers

• Analyzing and Correcting Benevolence Bias in Large Language Models [25.94213078450494]
  Large language models (LLMs) are increasingly used as stand-ins for human respondents.<n>Here we identify and measure benevolence bias, a small but consistent tendency for aligned LLMs to lean toward the kinder, safer, more socially approved answer on value-laden survey questions.<n>We find that the bias is a stable model property, not a quirk of any one system.
  arXiv  Detail & Related papers  (2026-07-26T08:37:58Z)
• Scale Can't Overcome Pragmatics: The Impact of Reporting Bias on Vision-Language Reasoning [79.95774256444956]
  The lack of reasoning capabilities in Vision-Language Models has remained at the forefront of research discourse.<n>We investigate the data underlying the popular VLMs OpenCLIP, LLaVA-1.5 and Molmo through the lens of theories from pragmatics.
  arXiv  Detail & Related papers  (2026-02-26T18:54:06Z)
• Biases in the Blind Spot: Detecting What LLMs Fail to Mention [13.599496385950983]
  Large Language Models (LLMs) often provide chain-of-thought (CoT) reasoning traces that appear plausible, but may hide internal biases.<n>We introduce a fully automated, black-box pipeline for detecting task-specific unverbalized biases.
  arXiv  Detail & Related papers  (2026-02-10T18:59:56Z)
• Bias in the Picture: Benchmarking VLMs with Social-Cue News Images and LLM-as-Judge Assessment [8.451522319478512]
  We introduce a news-image benchmark consisting of 1,343 image-question pairs drawn from diverse outlets.<n>We evaluate a range of state-of-the-art VLMs and employ a large language model (LLM) as judge, with human verification.<n>Our findings show that: (i) visual context systematically shifts model outputs in open-ended settings; (ii) bias prevalence varies across attributes and models, with particularly high risk for gender and occupation; and (iii) higher faithfulness does not necessarily correspond to lower bias.
  arXiv  Detail & Related papers  (2025-09-24T00:33:58Z)
• Robustly Improving LLM Fairness in Realistic Settings via Interpretability [0.16843915833103415]
  Anti-bias prompts fail when realistic contextual details are introduced.<n>We find that adding realistic context such as company names, culture descriptions from public careers pages, and selective hiring constraints induces significant racial and gender biases.<n>Our internal bias mitigation identifies race and gender-correlated directions and applies affine concept editing at inference time.
  arXiv  Detail & Related papers  (2025-06-12T17:34:38Z)
• Surface Fairness, Deep Bias: A Comparative Study of Bias in Language Models [45.41676783204022]
  We investigate various proxy measures of bias in large language models (LLMs)<n>We find that evaluating models with pre-prompted personae on a multi-subject benchmark (MMLU) leads to negligible and mostly random differences in scores.<n>With the recent trend for LLM assistant memory and personalization, these problems open up from a different angle.
  arXiv  Detail & Related papers  (2025-06-12T08:47:40Z)
• Where Fact Ends and Fairness Begins: Redefining AI Bias Evaluation through Cognitive Biases [77.3489598315447]
  We argue that identifying the boundary between fact and fair is essential for meaningful fairness evaluation.<n>We introduce Fact-or-Fair, a benchmark with (i) objective queries aligned with descriptive, fact-based judgments, and (ii) subjective queries aligned with normative, fairness-based judgments.
  arXiv  Detail & Related papers  (2025-02-09T10:54:11Z)
• GenderBias-\emph{VL}: Benchmarking Gender Bias in Vision Language Models via Counterfactual Probing [72.0343083866144]
  This paper introduces the GenderBias-emphVL benchmark to evaluate occupation-related gender bias in Large Vision-Language Models. Using our benchmark, we extensively evaluate 15 commonly used open-source LVLMs and state-of-the-art commercial APIs. Our findings reveal widespread gender biases in existing LVLMs.
  arXiv  Detail & Related papers  (2024-06-30T05:55:15Z)
• Self-Debiasing Large Language Models: Zero-Shot Recognition and Reduction of Stereotypes [73.12947922129261]
  We leverage the zero-shot capabilities of large language models to reduce stereotyping. We show that self-debiasing can significantly reduce the degree of stereotyping across nine different social groups. We hope this work opens inquiry into other zero-shot techniques for bias mitigation.
  arXiv  Detail & Related papers  (2024-02-03T01:40:11Z)
• GPTBIAS: A Comprehensive Framework for Evaluating Bias in Large Language Models [83.30078426829627]
  Large language models (LLMs) have gained popularity and are being widely adopted by a large user community. The existing evaluation methods have many constraints, and their results exhibit a limited degree of interpretability. We propose a bias evaluation framework named GPTBIAS that leverages the high performance of LLMs to assess bias in models.
  arXiv  Detail & Related papers  (2023-12-11T12:02:14Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.