# Moral Sycophancy in Vision Language Models

arXiv: https://arxiv.org/abs/2602.08311 · HTML: https://mlsanity.com/paper/2602.08311/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in VLM benchmarks and visual reasoning, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 15 of its 27 references that are in the map.
- First posted on arXiv 9 Feb 2026.

## Placed in

- Multimodal LLMs / vision-language models / [VLM benchmarks and visual reasoning](https://mlsanity.com/mllm/vlm-benchmarks-and-visual-reasoning/)

## Builds on (most cited first, 15 of 15)

- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution](https://mlsanity.com/paper/2409.12191/) (2024)
- [From Yes-Men to Truth-Tellers: Addressing Sycophancy in Large Language Models with Pinpoint Tuning](https://mlsanity.com/paper/2409.01658/) (2024)
- [MoralBench: Moral Evaluation of LLMs](https://mlsanity.com/paper/2406.04428/) (2024)
- [Quantifying and Mitigating Unimodal Biases in Multimodal Large Language Models: A Causal Perspective](https://mlsanity.com/paper/2403.18346/) (2024)
- [How Trustworthy are Open-Source LLMs? An Assessment under Malicious Demonstrations Shows their Vulnerabilities](https://mlsanity.com/paper/2311.09447/) (2023)
- [Looking Beyond Text: Reducing Language bias in Large Vision-Language Models via Multimodal Dual-Attention and Soft-Image Guidance](https://mlsanity.com/paper/2411.14279/) (2024)
- [textttModSCAN: Measuring Stereotypical Bias in Large Vision-Language Models from Vision and Language Modalities](https://mlsanity.com/paper/2410.06967/) (2024)
- [Have the VLMs Lost Confidence? A Study of Sycophancy in VLMs](https://mlsanity.com/paper/2410.11302/) (2024)
- [MORALISE: A Structured Benchmark for Moral Alignment in Visual Language Models](https://mlsanity.com/paper/2505.14728/) (2025)
- [MM-MoralBench: A MultiModal Moral Evaluation Benchmark for Large Vision-Language Models](https://mlsanity.com/paper/2412.20718/) (2024)
- [Measuring Agreeableness Bias in Multimodal Models](https://mlsanity.com/paper/2408.09111/) (2024)
- [FairPIVARA: Reducing and Assessing Biases in CLIP-Based Multimodal Models](https://mlsanity.com/paper/2409.19474/) (2024)
- [Sycophancy Claims about Language Models: The Missing Human-in-the-Loop](https://mlsanity.com/paper/2512.00656/) (2025)
- [PENDULUM: A Benchmark for Assessing Sycophancy in Multimodal Large Language Models](https://mlsanity.com/paper/2512.19350/) (2025)

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

- [SWAY: A Counterfactual Computational Linguistic Approach to Measuring and Mitigating Sycophancy](https://mlsanity.com/paper/2604.02423/) (2026)
- [When Helpfulness Becomes Sycophancy: Sycophancy is a Boundary Failure Between Social Alignment and Epistemic Integrity in Large Language Models](https://mlsanity.com/paper/2605.05403/) (2026)
- [Normative Robustness as a Frontier for Non-Verifiable Reasoning in LLMs](https://mlsanity.com/paper/2606.12731/) (2026)

## Abstract (arXiv)

Sycophancy in Vision-Language Models (VLMs) refers to their tendency to align with user opinions, often at the expense of moral or factual accuracy. While prior studies have explored sycophantic behavior in general contexts, its impact on morally grounded visual decision-making remains insufficiently understood. To address this gap, we present the first systematic study of moral sycophancy in VLMs, analyzing ten widely-used models on the Moralise and M^3oralBench datasets under explicit user disagreement. Our results reveal that VLMs frequently produce morally incorrect follow-up responses even when their initial judgments are correct, and exhibit a consistent asymmetry: models are more likely to shift from morally right to morally wrong judgments than the reverse when exposed to user-induced bias. Follow-up prompts generally degrade performance on Moralise, while yielding mixed or even improved accuracy on M^3oralBench, highlighting dataset-dependent differences in moral robustness. Evaluation using Error Introduction Rate (EIR) and Error Correction Rate (ECR) reveals a clear trade-off: models with stronger error-correction capabilities tend to introduce more reasoning errors, whereas more conservative models minimize errors but exhibit limited ability to self-correct. Finally, initial contexts with a morally right stance elicit stronger sycophantic behavior, emphasizing the vulnerability of VLMs to moral influence and the need for principled strategies to improve ethical consistency and robustness in multimodal AI systems.