# SAVeS: Steering Safety Judgments in Vision-Language Models via Semantic Cues

arXiv: https://arxiv.org/abs/2603.19092 · HTML: https://mlsanity.com/paper/2603.19092/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in MLLM jailbreaks and adversarial robustness, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 27 of its 28 references that are in the map.
- First posted on arXiv 19 Mar 2026.

## Placed in

- Multimodal LLMs / vision-language models / [MLLM jailbreaks and adversarial robustness](https://mlsanity.com/mllm/mllm-jailbreaks-and-adversarial-robustness/)

## Builds on (most cited first, 27 of 27)

- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://mlsanity.com/paper/2301.12597/) (2023)
- [Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context](https://mlsanity.com/paper/2403.05530/) (2024)
- [MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Models](https://mlsanity.com/paper/2304.10592/) (2023)
- [PaLM-E: An Embodied Multimodal Language Model](https://mlsanity.com/paper/2303.03378/) (2023)
- [Mitigating Object Hallucinations in Large Vision-Language Models through Visual Contrastive Decoding](https://mlsanity.com/paper/2311.16922/) (2023)
- [Visual ChatGPT: Talking, Drawing and Editing with Visual Foundation Models](https://mlsanity.com/paper/2303.04671/) (2023)
- [Safe RLHF: Safe Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2310.12773/) (2023)
- [Hallucination of Multimodal Large Language Models: A Survey](https://mlsanity.com/paper/2404.18930/) (2024)
- [MM-SafetyBench: A Benchmark for Safety Evaluation of Multimodal Large Language Models](https://mlsanity.com/paper/2311.17600/) (2023)
- [SafeAgentBench: A Benchmark for Safe Task Planning of Embodied LLM Agents](https://mlsanity.com/paper/2412.13178/) (2024)
- [VLSBench: Unveiling Visual Leakage in Multimodal Safety](https://mlsanity.com/paper/2411.19939/) (2024)
- [Multimodal Situational Safety](https://mlsanity.com/paper/2410.06172/) (2024)
- [IS-Bench: Evaluating Interactive Safety of VLM-Driven Embodied Agents in Daily Household Tasks](https://mlsanity.com/paper/2506.16402/) (2025)
- [AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions](https://mlsanity.com/paper/2506.14697/) (2025)
- [IDEATOR: Jailbreaking and Benchmarking Large Vision-Language Models Using Themselves](https://mlsanity.com/paper/2411.00827/) (2024)
- [Treble Counterfactual VLMs: A Causal Approach to Hallucination](https://mlsanity.com/paper/2503.06169/) (2025)
- [Safety Guardrails for LLM-Enabled Robots](https://mlsanity.com/paper/2503.07885/) (2025)
- [Safe RLHF-V: Safe Reinforcement Learning from Multi-modal Human Feedback](https://mlsanity.com/paper/2503.17682/) (2025)
- [Visual Contextual Attack: Jailbreaking MLLMs with Image-Driven Context Injection](https://mlsanity.com/paper/2507.02844/) (2025)
- [Unveiling Typographic Deceptions: Insights of the Typographic Vulnerability in Large Vision-Language Model](https://mlsanity.com/paper/2402.19150/) (2024)
- [DASH: Detection and Assessment of Systematic Hallucinations of VLMs](https://mlsanity.com/paper/2503.23573/) (2025)
- [Don't Let Your Robot be Harmful: Responsible Robotic Manipulation via Safety-as-Policy](https://mlsanity.com/paper/2411.18289/) (2024)
- [EAGLE: Enhanced Visual Grounding Minimizes Hallucinations in Instructional Multimodal Models](https://mlsanity.com/paper/2501.02699/) (2025)
- [Multimodal Safety Evaluation in Generative Agent Social Simulations](https://mlsanity.com/paper/2510.07709/) (2025)
- [Vision Language Models Map Logos to Text via Semantic Entanglement in the Visual Projector](https://mlsanity.com/paper/2510.12287/) (2025)

## Built on by (most cited first, 4 of 4)

- [When Think-with-Image Meets Safety: What Determines Multimodal Jailbreak Robustness?](https://mlsanity.com/paper/2605.27932/) (2026)
- [SGA: Plug&Play Geometric Verification for Educational Video Synthesis](https://mlsanity.com/paper/2607.18116/) (2026)
- [When Safety Overrides Vision: Exploring Dynamics between Vision Influence and Safety Alignment in Vision-Language Models](https://mlsanity.com/paper/2608.18628/) (2026)
- [The Uncontrolled Variable: Vision-Language Model Refusal Responds to Image Presence in Ways Risk Cannot Explain](https://mlsanity.com/paper/2609.26174/) (2026)

## Abstract (arXiv)

Vision-language models (VLMs) are increasingly deployed in real-world and embodied settings where safety decisions depend on visual context. However, it remains unclear which visual evidence drives these judgments. We study whether multimodal safety behavior in VLMs can be steered by simple semantic cues. We introduce a semantic steering framework that applies controlled textual, visual, and cognitive interventions without changing the underlying scene content. To evaluate these effects, we propose SAVeS, a benchmark for situational safety under semantic cues, together with an evaluation protocol that separates behavioral refusal, grounded safety reasoning, and false refusals. Experiments across multiple VLMs and an additional state-of-the-art benchmark show that safety decisions are highly sensitive to semantic cues, indicating reliance on learned visual-linguistic associations rather than grounded visual understanding. We further demonstrate that automated steering pipelines can exploit these mechanisms, highlighting a potential vulnerability in multimodal safety systems.