Safe Semantics, Unsafe Interpretations: Tackling Implicit Reasoning Safety in Large Vision-Language Models
On the map: LLM reasoning → Reasoning for agents, planning and code
Abstract (arXiv)
Large Vision-Language Models face growing safety challenges with multimodal inputs. This paper introduces the concept of Implicit Reasoning Safety, a vulnerability in LVLMs. Benign combined inputs trigger unsafe LVLM outputs due to flawed or hidden reasoning. To showcase this, we developed Safe Semantics, Unsafe Interpretations, the first dataset for this critical issue. Our demonstrations show that even simple In-Context Learning with SSUI significantly mitigates these implicit multimodal threats, underscoring the urgent need to improve cross-modal implicit reasoning.
Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning. Builds on 6 of its 14 references that are in the map. First posted on arXiv 12 Aug 2025.
Builds on
6 of 14 references.
References · 6
- 2025Qwen2.5-VL Technical Reportcited by 5,587
- 2023Qwen Technical Reportcited by 3,344
- 2024Gemini 1.5: Unlocking multimodal understanding across millions of tokens of contextcited by 3,224
- 2022A Survey on In-context Learningcited by 721
- 2025Mimic In-Context Learning for Multimodal Taskscited by 12
- 2020Egoshots, an ego-vision life-logging dataset and semantic fidelity metric to evaluate diversity in image captioning modelscited by 8
Built on by
Citing papers per year: 2025: 2; 2026: 1.
3 papers.
- 2025Aetheria: A multimodal interpretable content safety framework based on multi-agent debate and collaborationcited by 2
- 2025Visual Attention Reasoning via Hierarchical Search and Self-Verificationcited by 0
- 2026Distributed Implicit Harm: A Compositional Safety Blind Spot in MLLM-Based Video Moderationcited by 0
- 2026Distributed Implicit Harm: A Compositional Safety Blind Spot in MLLM-Based Video Moderation31 Aug 2026
- 2025Aetheria: A multimodal interpretable content safety framework based on multi-agent debate and collaboration2 Dec 2025
- 2025Visual Attention Reasoning via Hierarchical Search and Self-Verification21 Oct 2025
Similar, not cited
Close in wording, no citation either way
- 2025When Safe Unimodal Inputs Collide: Optimizing Reasoning Chains for Cross-Modal Safety in Multimodal Large Language Models
- 2026Two Birds, One Projection: Harmonizing Safety and Utility in LVLMs via Inference-time Feature Projection
- 2025Rethinking Bottlenecks in Safety Fine-Tuning of Vision Language Models
- 2025VLSU: Mapping the Limits of Joint Multimodal Understanding for AI Safety
- 2025Think-Reflect-Revise: A Policy-Guided Reflective Framework for Safety Alignment in Large Vision Language Models
- 2025SIA: Enhancing Safety via Intent Awareness for Vision-Language Models
- 2024ETA: Evaluating Then Aligning Safety of Vision Language Models at Inference Time
- 2025MSR-Align: Policy-Grounded Multimodal Alignment for Safety-Aware Reasoning in Vision-Language Models