Pith. sign in

REVIEW 5 cited by

VLM-Guard: Safeguarding Vision-Language Models via Fulfilling Safety Alignment Gap

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2502.10486 v1 pith:XF536XGY submitted 2025-02-14 cs.CR cs.AIcs.CV

VLM-Guard: Safeguarding Vision-Language Models via Fulfilling Safety Alignment Gap

classification cs.CR cs.AIcs.CV
keywords safetyalignmentvlm-guardvlmscomponentfulfillingllmsmodality
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

The emergence of vision language models (VLMs) comes with increased safety concerns, as the incorporation of multiple modalities heightens vulnerability to attacks. Although VLMs can be built upon LLMs that have textual safety alignment, it is easily undermined when the vision modality is integrated. We attribute this safety challenge to the modality gap, a separation of image and text in the shared representation space, which blurs the distinction between harmful and harmless queries that is evident in LLMs but weakened in VLMs. To avoid safety decay and fulfill the safety alignment gap, we propose VLM-Guard, an inference-time intervention strategy that leverages the LLM component of a VLM as supervision for the safety alignment of the VLM. VLM-Guard projects the representations of VLM into the subspace that is orthogonal to the safety steering direction that is extracted from the safety-aligned LLM. Experimental results on three malicious instruction settings show the effectiveness of VLM-Guard in safeguarding VLM and fulfilling the safety alignment gap between VLM and its LLM component.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. V-DEAL: Diagnosing Video Safety De-Calibration as an Understanding-Refusal Coupling Failure

    cs.AI 2026-07 conditional novelty 6.0

    Video LLMs understand harmful content but activate weaker refusal signals when the query is benign; prompt realignment reduces attack success from ~48% to ~1%.

  2. Safety Geometry Collapse in Multimodal LLMs and Adaptive Drift Correction

    cs.AI 2026-05 unverdicted novelty 5.0

    Multimodal LLMs suffer Safety Geometry Collapse from modality-induced drift that reduces refusal separability; ReGap corrects drift at inference time using self-rectification signals to restore safety without retraining.

  3. WARD: Adversarially Robust Defense of Web Agents Against Prompt Injections

    cs.CR 2026-05 unverdicted novelty 5.0

    WARD is a guard model trained on 177K web samples and adversarially hardened via attacker-guard co-evolution to achieve high recall on prompt injections with low false positives and no added latency.

  4. Reading Between the Pixels: Linking Text-Image Embedding Alignment to Typographic Attack Success on Vision-Language Models

    cs.CV 2026-04 unverdicted novelty 5.0

    Text-image embedding distance negatively correlates with typographic attack success rates (r = -0.71 to -0.93) on VLMs, with font size and image degradations strongly modulating effectiveness.

  5. SALLIE: Safeguarding Against Latent Language & Image Exploits

    cs.CR 2026-04 unverdicted novelty 5.0

    SALLIE detects jailbreaks in text and vision-language models by extracting residual stream activations, scoring maliciousness per layer with k-NN, and ensembling predictions, outperforming baselines on multiple datasets.