# Once Correct, Still Wrong: Counterfactual Hallucination in Multilingual Vision-Language Models

arXiv: https://arxiv.org/abs/2602.05437 · HTML: https://mlsanity.com/paper/2602.05437/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in Hallucination mitigation in LVLMs, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 14 of its 32 references that are in the map.
- First posted on arXiv 5 Feb 2026.

## Placed in

- Multimodal LLMs / vision-language models / [Hallucination mitigation in LVLMs](https://mlsanity.com/mllm/hallucination-mitigation-in-lvlms/)

## Builds on (most cited first, 14 of 14)

- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Large Language Models are Zero-Shot Reasoners](https://mlsanity.com/paper/2205.11916/) (2022)
- [Can language models learn from explanations in context?](https://mlsanity.com/paper/2204.02329/) (2022)
- [CVQA: Culturally-diverse Multilingual Visual Question Answering Benchmark](https://mlsanity.com/paper/2406.05967/) (2024)
- [Negative Object Presence Evaluation (NOPE) to Measure Object Hallucination in Vision-Language Models](https://mlsanity.com/paper/2310.05338/) (2023)
- [Mitigating Hallucination in Visual Language Models with Visual Supervision](https://mlsanity.com/paper/2311.16479/) (2023)
- [AraDiCE: Benchmarks for Dialectal and Cultural Capabilities in LLMs](https://mlsanity.com/paper/2409.11404/) (2024)
- [PhD: A ChatGPT-Prompted Visual hallucination Evaluation Dataset](https://mlsanity.com/paper/2403.11116/) (2024)
- [Treble Counterfactual VLMs: A Causal Approach to Hallucination](https://mlsanity.com/paper/2503.06169/) (2025)
- [A Survey of Multimodal Hallucination Evaluation and Detection](https://mlsanity.com/paper/2507.19024/) (2025)
- [Behind the Magic, MERLIM: Multi-modal Evaluation Benchmark for Large Image-Language Models](https://mlsanity.com/paper/2312.02219/) (2023)
- [Dallah: A Dialect-Aware Multimodal Large Language Model for Arabic](https://mlsanity.com/paper/2407.18129/) (2024)
- [The Landscape of Arabic Large Language Models (ALLMs): A New Era for Arabic Language Technology](https://mlsanity.com/paper/2506.01340/) (2025)
- [AIN: The Arabic INclusive Large Multimodal Model](https://mlsanity.com/paper/2502.00094/) (2025)

## Built on by (most cited first, 2 of 2)

- [Said Aloud, Read Different: Cross-Modal Instability in Multimodal Models](https://mlsanity.com/paper/2608.27135/) (2026)
- [When Words Speak Louder than Images: Towards Understanding Language Bias in Vision-Language Models](https://mlsanity.com/paper/2609.35272/) (2026)

## Abstract (arXiv)

Vision-language models (VLMs) can achieve high accuracy while still accepting culturally plausible but visually incorrect interpretations. Existing hallucination benchmarks rarely test this failure mode, particularly outside Western contexts and English. We introduce M2CQA, a culturally grounded multimodal benchmark built from images spanning 17 MENA countries, paired with contrastive true and counterfactual statements in English, Arabic, and its dialects. To isolate hallucination beyond raw accuracy, we propose the CounterFactual Hallucination Rate (CFHR), which measures counterfactual acceptance conditioned on correctly answering the true statement. Evaluating state-of-the-art VLMs under multiple prompting strategies, we find that CFHR rises sharply in Arabic, especially in dialects, even when true-statement accuracy remains high. Moreover, reasoning-first prompting consistently increases counterfactual hallucination, while answering before justifying improves robustness. We make the dataset publicly available for the community (https://huggingface.co/datasets/QCRI/M2CQA)).