# Bias in the Picture: Benchmarking VLMs with Social-Cue News Images and LLM-as-Judge Assessment

arXiv: https://arxiv.org/abs/2509.19659 · HTML: https://mlsanity.com/paper/2509.19659/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Social bias and fairness in VLMs, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 19 of its 28 references that are in the map.
- First posted on arXiv 24 Sep 2025.

## Placed in

- Multimodal LLMs / vision-language models / [Social bias and fairness in VLMs](https://mlsanity.com/mllm/social-bias-and-fairness-in-vlms/)

## Builds on (most cited first, 19 of 19)

- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks](https://mlsanity.com/paper/2312.14238/) (2023)
- [ChatGLM: A Family of Large Language Models from GLM-130B to GLM-4 All Tools](https://mlsanity.com/paper/2406.12793/) (2024)
- [Vision-Language Models for Vision Tasks: A Survey](https://mlsanity.com/paper/2304.00685/) (2023)
- [Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scaling](https://mlsanity.com/paper/2501.17811/) (2025)
- [Phi-4 Technical Report](https://mlsanity.com/paper/2412.08905/) (2024)
- [PaliGemma: A versatile 3B VLM for transfer](https://mlsanity.com/paper/2407.07726/) (2024)
- [DecodingTrust: A Comprehensive Assessment of Trustworthiness in GPT Models](https://mlsanity.com/paper/2306.11698/) (2023)
- [Multimodal Large Language Models: A Survey](https://mlsanity.com/paper/2311.13165/) (2023)
- [Evaluating CLIP: Towards Characterization of Broader Capabilities and Downstream Implications](https://mlsanity.com/paper/2108.02818/) (2021)
- [Examining Gender and Racial Bias in Large Vision-Language Models Using a Novel Dataset of Parallel Images](https://mlsanity.com/paper/2402.05779/) (2024)
- [SocialCounterfactuals: Probing and Mitigating Intersectional Social Biases in Vision-Language Models with Counterfactual Examples](https://mlsanity.com/paper/2312.00825/) (2023)
- [MultiTrust: A Comprehensive Benchmark Towards Trustworthy Multimodal Large Language Models](https://mlsanity.com/paper/2406.07057/) (2024)
- [Worst of Both Worlds: Biases Compound in Pre-trained Vision-and-Language Models](https://mlsanity.com/paper/2104.08666/) (2021)
- [VisoGender: A dataset for benchmarking gender bias in image-text pronoun resolution](https://mlsanity.com/paper/2306.12424/) (2023)
- [GenderBias-VL: Benchmarking Gender Bias in Vision Language Models via Counterfactual Probing](https://mlsanity.com/paper/2407.00600/) (2024)
- [Survey of Social Bias in Vision-Language Models](https://mlsanity.com/paper/2309.14381/) (2023)
- [Enhancing Advanced Visual Reasoning Ability of Large Language Models](https://mlsanity.com/paper/2409.13980/) (2024)
- [The Bias of Harmful Label Associations in Vision-Language Models](https://mlsanity.com/paper/2402.07329/) (2024)

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

- [BioPro: Towards Difference-Aware Gender Fairness for Vision-Language Models](https://mlsanity.com/paper/2512.00807/) (2025)
- [Cultural Counterfactuals: Evaluating Cultural Biases in Large Vision-Language Models with Counterfactual Examples](https://mlsanity.com/paper/2603.02370/) (2026)
- [LLM-as-Judge for Semantic Judging of Powerline Segmentation in UAV Inspection](https://mlsanity.com/paper/2604.05371/) (2026)

## Abstract (arXiv)

Large vision-language models (VLMs) can jointly interpret images and text, but they are also prone to absorbing and reproducing harmful social stereotypes when visual cues such as age, gender, race, clothing, or occupation are present. To investigate these risks, we introduce a news-image benchmark consisting of 1,343 image-question pairs drawn from diverse outlets, which we annotated with ground-truth answers and demographic attributes (age, gender, race, occupation, and sports). We evaluate a range of state-of-the-art VLMs and employ a large language model (LLM) as judge, with human verification. 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. We release the benchmark prompts, evaluation rubric, and code to support reproducible and fairness-aware multimodal assessment.