V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure | Paper Citation Record

Observation 523e1f8d-aaa2-4679-8667-bb6b4fc3243e · outbound

This paper cites Qwen3-vl technical report,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Qwen3-vl technical report,

Reference 1

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:24.265213Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0cec7b8c-3c74-45d4-8161-880fdb5788a6 · outbound

This paper cites InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency

Reference 2

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:24.414167Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ff3267ed-70a9-484c-bfc9-b1e7b5a9304d · outbound

This paper cites InternVideo2.5: Empowering Video MLLMs with Long and Rich Context Modeling.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure InternVideo2.5: Empowering Video MLLMs with Long and Rich Context Modeling

Reference 3

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:24.495566Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f1501e31-11e0-4751-b687-4403fe63d4a1 · outbound

This paper cites Vision-language models for vision tasks: A survey,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Vision-language models for vision tasks: A survey,

Reference 4

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:24.732526Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0e45cb66-a857-409a-996d-155224d02cf1 · outbound

This paper cites Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution

Reference 5

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:24.901910Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d54a7677-0c02-45c0-a740-7977675c926f · outbound

This paper cites Safety of Multimodal Large Language Models on Images and Texts.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Safety of Multimodal Large Language Models on Images and Texts

Reference 6

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:25.032757Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a4b7d2bc-1709-4d4c-a6fd-1c95fbc5f6b3 · outbound

This paper cites Jailbreak Attacks and Defenses against Multimodal Generative Models: A Survey.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Jailbreak Attacks and Defenses against Multimodal Generative Models: A Survey

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:25.193531Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4f1ed196-4b14-46ee-ab10-788259ce4909 · outbound

This paper cites Vlsbench: Unveiling visual leakage in multimodal safety,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Vlsbench: Unveiling visual leakage in multimodal safety,

Reference 8

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:25.333115Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4f03dbc6-603b-4ded-b8bf-e21a660b6bc2 · outbound

This paper cites Video-safetybench: A benchmark for safety evaluation of video lvlms,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Video-safetybench: A benchmark for safety evaluation of video lvlms,

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:25.472715Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0a720c0d-0f09-49df-ac44-d317cfab5980 · outbound

This paper cites Sea: Low-resource safety alignment for multimodal large language models via synthetic embeddings,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Sea: Low-resource safety alignment for multimodal large language models via synthetic embeddings,

Reference 10

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:25.646133Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e8ace07f-185f-4fa4-83a5-e562128e3cc6 · outbound

This paper cites Figstep: Jail- breaking large vision-language models via typographic visual prompts,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Figstep: Jail- breaking large vision-language models via typographic visual prompts,

Reference 11

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:25.779852Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8d8a2ab6-d404-41d3-971f-ace533de2ab2 · outbound

This paper cites Visual-RolePlay: Universal Jailbreak Attack on MultiModal Large Language Models via Role-playing Image Character.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Visual-RolePlay: Universal Jailbreak Attack on MultiModal Large Language Models via Role-playing Image Character

Reference 12

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:25.911168Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 40ece37e-0c78-405a-99c1-3cbaea5b3982 · outbound

This paper cites Ideator: Jailbreaking and benchmarking large vision-language models using them- selves,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Ideator: Jailbreaking and benchmarking large vision-language models using them- selves,

Reference 13

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:26.010574Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1291fbc4-d084-4448-9111-d280c0f35c09 · outbound

This paper cites JailBreakV: A Benchmark for Assessing the Robustness of MultiModal Large Language Models against Jailbreak Attacks.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure JailBreakV: A Benchmark for Assessing the Robustness of MultiModal Large Language Models against Jailbreak Attacks

Reference 14

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:26.121407Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 04e745c3-3952-4397-a67f-fd6de91ff9a3 · outbound

This paper cites $\textit{MMJ-Bench}$: A Comprehensive Study on Jailbreak Attacks and Defenses for Multimodal Large Language Models.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure $\textit{MMJ-Bench}$: A Comprehensive Study on Jailbreak Attacks and Defenses for Multimodal Large Language Models

Reference 15

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:26.274455Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8589873e-2a79-4293-8974-7c5712661bb7 · outbound

This paper cites Omni-SafetyBench: A benchmark for safety evaluation of audio-visual large language models,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Omni-SafetyBench: A benchmark for safety evaluation of audio-visual large language models,

Reference 16

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:26.442624Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0799568d-d601-4bdb-b40f-01e8d412a9d1 · outbound

This paper cites Videostir: Understanding long videos via spatio-temporally structured and intent-aware rag,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Videostir: Understanding long videos via spatio-temporally structured and intent-aware rag,

Reference 17

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:26.575880Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 24f3724e-cae4-4827-bc03-a84c4437508a · outbound

This paper cites CamReasoner: Reinforcing Camera Movement Understanding via Structured Spatial Reasoning.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure CamReasoner: Reinforcing Camera Movement Understanding via Structured Spatial Reasoning

Reference 18

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:26.724359Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 36976e9f-b72e-4da3-87b6-8ad0f773f60d · outbound

This paper cites Mm-safetybench: A benchmark for safety evaluation of multimodal large language models,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Mm-safetybench: A benchmark for safety evaluation of multimodal large language models,

Reference 19

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:26.846948Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a5e9b5bb-6654-4fe7-9342-345395041551 · outbound

This paper cites Xstest: A test suite for identifying exaggerated safety behaviours in large language models,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Xstest: A test suite for identifying exaggerated safety behaviours in large language models,

Reference 20

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:27.008533Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c24710b5-bd32-4c48-a740-b1448f5570fc · outbound

This paper cites HiddenGuard: Fine-Grained Safe Generation with Specialized Representation Router.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure HiddenGuard: Fine-Grained Safe Generation with Specialized Representation Router

Reference 21

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:27.163780Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation bbdb4910-4e60-4fde-ac4e-5eb904998a4c · outbound

This paper cites You Know What I'm Saying: Jailbreak Attack via Implicit Reference.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure You Know What I'm Saying: Jailbreak Attack via Implicit Reference

Reference 22

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:27.312864Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ea9bd991-09de-47c9-9bb2-4d0721a37be1 · outbound

This paper cites Spa-vl: A comprehensive safety preference alignment dataset for vision language models,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Spa-vl: A comprehensive safety preference alignment dataset for vision language models,

Reference 23

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:27.461462Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4b465fa7-55d3-4f53-aa60-62705d4cfda4 · outbound

This paper cites Msr-align: Policy-grounded multimodal alignment for safety-aware reasoning in vision-language models,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Msr-align: Policy-grounded multimodal alignment for safety-aware reasoning in vision-language models,

Reference 24

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:27.575849Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7e02f988-e635-400b-a83d-0e25b20f020b · outbound

This paper cites Safety Fine-Tuning at (Almost) No Cost: A Baseline for Vision Large Language Models.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Safety Fine-Tuning at (Almost) No Cost: A Baseline for Vision Large Language Models

Reference 25

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:27.722142Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 310881d8-b230-4136-9ee3-2b7a756e426e · outbound

This paper cites Adashield: Safeguarding multimodal large language models from structure-based attack via adaptive shield prompting,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Adashield: Safeguarding multimodal large language models from structure-based attack via adaptive shield prompting,

Reference 26

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:27.870185Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3b66fdb5-54da-4621-91ef-69c857226bd6 · outbound

This paper cites Bluesuffix: Reinforced blue teaming for vision-language models against jailbreak attacks,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Bluesuffix: Reinforced blue teaming for vision-language models against jailbreak attacks,

Reference 27

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:28.009260Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 01f990d7-6f45-4a36-bd9c-700efb04c40f · outbound

This paper cites E2AT: Multimodal Jailbreak Defense via Dynamic Joint Optimization for Multimodal Large Language Models,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure E2AT: Multimodal Jailbreak Defense via Dynamic Joint Optimization for Multimodal Large Language Models,

Reference 28

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:28.133850Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 07f65600-5646-4f35-ae3c-5cc7e167741d · outbound

This paper cites Eta: Evaluating then aligning safety of vision language models at inference time,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Eta: Evaluating then aligning safety of vision language models at inference time,

Reference 29

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:28.293786Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 226d2725-39a9-453a-a544-0f92bc29216e · outbound

This paper cites Safedecoding: Defending against jailbreak attacks via safety-aware decoding,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Safedecoding: Defending against jailbreak attacks via safety-aware decoding,

Reference 30

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:28.455095Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 74c85e18-a11e-4e1a-a77f-4188d01526cf · outbound

This paper cites Defending multimodal backdoored models by repulsive visual prompt tuning,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Defending multimodal backdoored models by repulsive visual prompt tuning,

Reference 31

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:28.556600Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0041fd77-69c9-4866-858f-31066c132e23 · outbound

This paper cites Gated differentiable working memory for long-context language modeling,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Gated differentiable working memory for long-context language modeling,

Reference 32

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:28.634357Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation aac1f20a-73f5-42a2-bf28-0c476ef2a120 · outbound

This paper cites Test-time attention purification for backdoored large vision language models,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Test-time attention purification for backdoored large vision language models,

Reference 33

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:28.719510Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 81a69b56-690e-451c-95b5-8dec922b5162 · outbound

This paper cites Mrfd: Multi-region fusion decoding with self-consistency for mitigating hallucinations in lvlms,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Mrfd: Multi-region fusion decoding with self-consistency for mitigating hallucinations in lvlms,

Reference 34

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:28.799554Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 118b9f87-2a9e-433f-abf5-bfa77a4b287c · outbound

This paper cites TokenSwap: Backdoor Attack on the Compositional Understanding of Large Vision-Language Models.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure TokenSwap: Backdoor Attack on the Compositional Understanding of Large Vision-Language Models

Reference 35

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:28.879633Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b26c8ed6-704d-491b-972e-e90a15e84d13 · outbound

This paper cites Dimo-gui: Advancing test-time scaling in gui grounding via modality-aware visual reasoning,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Dimo-gui: Advancing test-time scaling in gui grounding via modality-aware visual reasoning,

Reference 36

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:28.977259Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 65a08c8c-3f88-4687-80e8-befe5ba759d4 · outbound

This paper cites Improving generaliz- ability and undetectability for targeted adversarial attacks on multimodal pre-trained models,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Improving generaliz- ability and undetectability for targeted adversarial attacks on multimodal pre-trained models,

Reference 37

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:29.061898Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation bf77cec6-6d0c-4de2-84f0-5d857f8d57aa · outbound

This paper cites Representation Engineering: A Top-Down Approach to AI Transparency.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Representation Engineering: A Top-Down Approach to AI Transparency

Reference 38

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:29.149242Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 802859be-1d11-4bcd-8a5e-615a765b77f2 · outbound

This paper cites Refusal in language models is mediated by a single direction,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Refusal in language models is mediated by a single direction,

Reference 39

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:29.228401Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7182405a-2387-43d1-9069-b4d8d7b7f80c · outbound

This paper cites Guardreasoner-vl: Safeguarding vlms via reinforced reasoning,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Guardreasoner-vl: Safeguarding vlms via reinforced reasoning,

Reference 40

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:29.306894Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation fba53fc4-82a2-42ab-8546-6cd5b542d37f · outbound

This paper cites Spot Risks Before Speaking! Unraveling Safety Attention Heads in Large Vision-Language Models.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Spot Risks Before Speaking! Unraveling Safety Attention Heads in Large Vision-Language Models

Reference 41

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:29.396060Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2241f3d5-bfa8-4ef7-a6fa-57d22577eecf · outbound

This paper cites Jail- breaklens: Visual analysis of jailbreak attacks against large language models,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Jail- breaklens: Visual analysis of jailbreak attacks against large language models,

Reference 42

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:29.481321Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7391216b-c470-4022-8132-13e4d62df029 · outbound

This paper cites "not aligned.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure "not aligned

Reference 43

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:29.581759Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4b4b0dab-6a3f-4a77-a3dd-70cbd4b3efb0 · outbound

This paper cites Focusing by Contrastive Attention: Enhancing VLMs' Visual Reasoning.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Focusing by Contrastive Attention: Enhancing VLMs' Visual Reasoning

Reference 44

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:29.666893Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation efbcb1ad-39ac-4840-b030-692cdfce5d57 · outbound

This paper cites Effective and Efficient Adversarial Detection for Vision-Language Models via A Single Vector.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Effective and Efficient Adversarial Detection for Vision-Language Models via A Single Vector

Reference 45

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:29.745621Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c89cbb58-65f9-46d4-a841-f7f94bb71fb4 · outbound

This paper cites VLM-Guard: Safeguarding Vision-Language Models via Fulfilling Safety Alignment Gap.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure VLM-Guard: Safeguarding Vision-Language Models via Fulfilling Safety Alignment Gap

Reference 46

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:29.831115Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9dd1bd8a-8244-4790-afa5-b31096496715 · outbound

This paper cites Safety Subspaces are Not Linearly Distinct: A Fine-Tuning Case Study,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Safety Subspaces are Not Linearly Distinct: A Fine-Tuning Case Study,

Reference 47

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:29.961493Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5b3c30c2-be6d-4e5c-80de-9472927ce71b · outbound

This paper cites HiddenDetect: Detecting Jailbreak Attacks against Large Vision-Language Models via Monitoring Hidden States.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure HiddenDetect: Detecting Jailbreak Attacks against Large Vision-Language Models via Monitoring Hidden States

Reference 48

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:30.045950Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2cb625d7-3ab6-4ba9-9b7f-6bca1a30cb20 · outbound

This paper cites Brainvis: Exploring the bridge between brain and visual signals via image reconstruction,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Brainvis: Exploring the bridge between brain and visual signals via image reconstruction,

Reference 49

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:30.102859Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ff7c5261-a453-457f-8426-f2b941488006 · outbound

This paper cites Tuning vision-language models with candidate labels by prompt alignment,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Tuning vision-language models with candidate labels by prompt alignment,

Reference 50

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:30.217304Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation aa737389-de1b-479e-8a3e-c0a65151bfb0 · outbound

This paper cites Refineshot: Rethinking cinematography understanding with foundational skill evaluation,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Refineshot: Rethinking cinematography understanding with foundational skill evaluation,

Reference 51

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:30.354643Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2707110f-43c0-40e8-9446-3278e2f9779a · outbound

This paper cites What should a streaming video model remember?.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure What should a streaming video model remember?

Reference 52

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:30.430349Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation fc9c40da-8dca-4354-8b8a-9e8f9e7a5591 · outbound

This paper cites VLSU: Mapping the Limits of Joint Multimodal Understanding for AI Safety,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure VLSU: Mapping the Limits of Joint Multimodal Understanding for AI Safety,

Reference 53

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:30.536722Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 58db01ab-fba1-4b02-b875-2df991562a9e · outbound

This paper cites Vistawise: Building cost-effective agent with cross-modal knowledge graph for minecraft,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Vistawise: Building cost-effective agent with cross-modal knowledge graph for minecraft,

Reference 54

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:30.608162Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 42dea20b-383b-4903-befe-c2cf52b315fb · outbound

This paper cites Safe inputs but unsafe output: Benchmarking cross-modality safety alignment of large vision-language models,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Safe inputs but unsafe output: Benchmarking cross-modality safety alignment of large vision-language models,

Reference 55

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:30.700547Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ec738abf-5682-4d9e-9949-bc5c97e30644 · outbound

This paper cites Contextnav: Towards agentic multimodal in-context learning,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Contextnav: Towards agentic multimodal in-context learning,

Reference 56

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:30.785369Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a624a08a-09d9-4b82-ba71-dfb74e901cb5 · outbound

This paper cites Innate Reasoning is Not Enough: In-Context Learning Enhances Reasoning Large Language Models with Less Overthinking.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Innate Reasoning is Not Enough: In-Context Learning Enhances Reasoning Large Language Models with Less Overthinking

Reference 57

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:30.869177Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation acff3c2e-36fe-41ef-85c3-3f6dad372087 · outbound

This paper cites MSTS: A Multimodal Safety Test Suite for Vision-Language Models.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure MSTS: A Multimodal Safety Test Suite for Vision-Language Models

Reference 58

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:30.999477Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 329af082-73f1-40cf-927d-45246d7dec74 · outbound

This paper cites Enhanced partially relevant video retrieval through inter-and intra-sample analysis with coherence prediction,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Enhanced partially relevant video retrieval through inter-and intra-sample analysis with coherence prediction,

Reference 59

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:31.127006Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7ed234ea-30bd-4a88-aea4-96f441780839 · outbound

This paper cites Slang: New concept comprehension of large language models,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Slang: New concept comprehension of large language models,

Reference 60

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:31.257851Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8df1f85f-83e8-4bd7-a22c-e78884379a39 · outbound

This paper cites WaMo: Wavelet-Enhanced Multi-Frequency Trajectory Analysis for Fine-Grained Text-Motion Retrieval.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure WaMo: Wavelet-Enhanced Multi-Frequency Trajectory Analysis for Fine-Grained Text-Motion Retrieval

Reference 61

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:31.376601Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 530f78e2-97ce-4999-950d-9dd642de875f · outbound

This paper cites Sdr-gain: A high real-time occluded pedestrian pose completion method for autonomous driving,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Sdr-gain: A high real-time occluded pedestrian pose completion method for autonomous driving,

Reference 62

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:31.512899Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 81008fc1-aaf3-4e1e-88bf-16763bfc89cd · outbound

This paper cites MiniCPM-o 4.5: Towards Real-Time Full-Duplex Omni-Modal Interaction.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure MiniCPM-o 4.5: Towards Real-Time Full-Duplex Omni-Modal Interaction

Reference 63

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:31.649091Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 1f5e13f5-9c9b-4a31-8035-7a685dd5f467 · outbound

This paper cites Enhanced cross-modal 3d retrieval via tri-modal reconstruction,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Enhanced cross-modal 3d retrieval via tri-modal reconstruction,

Reference 64

Resolution: unresolved; no resolver link, observed 2026-08-01T08:23:31.730788Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 86184b48-d41f-42d1-8a98-0e156bc87929 · outbound

This paper cites Minicpm-v 4.5: Cooking efficient mllms via architecture, data, and training recipe,.

V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure Minicpm-v 4.5: Cooking efficient mllms via architecture, data, and training recipe,

Reference 65

Resolution: malformed identifier; no resolver link, observed 2026-08-01T08:23:31.950259Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Coverage vector

Reference resolution

Outbound references

Pith citing papers