Pith. sign in

REVIEW 8 cited by

Revisiting Backdoor Attacks against Large Vision-Language Models from Domain Shift

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 2406.18844 v4 pith:7D4762HG submitted 2024-06-27 cs.CV

classification cs.CV
keywords backdoorattackattacksdomaingeneralizabilityratedatadomains
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Instruction tuning enhances large vision-language models (LVLMs) but increases their vulnerability to backdoor attacks due to their open design. Unlike prior studies in static settings, this paper explores backdoor attacks in LVLM instruction tuning across mismatched training and testing domains. We introduce a new evaluation dimension, backdoor domain generalization, to assess attack robustness under visual and text domain shifts. Our findings reveal two insights: (1) backdoor generalizability improves when distinctive trigger patterns are independent of specific data domains or model architectures, and (2) the competitive interaction between trigger patterns and clean semantic regions, where guiding the model to predict triggers enhances attack generalizability. Based on these insights, we propose a multimodal attribution backdoor attack (MABA) that injects domain-agnostic triggers into critical areas using attributional interpretation. Experiments with OpenFlamingo, Blip-2, and Otter show that MABA significantly boosts the attack success rate of generalization by 36.4%, achieving a 97% success rate at a 0.2% poisoning rate. This study reveals limitations in current evaluations and highlights how enhanced backdoor generalizability poses a security threat to LVLMs, even without test data access.

Discussion (0). Sign in to comment.

Forward citations

Cited by 8 Pith papers

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

  1. TokenSwap: Backdoor Attack on the Compositional Understanding of Large Vision-Language Models

    cs.CV 2025-09 conditional novelty 6.0 of 10

    TokenSwap poisons LVLMs so that triggered images produce captions with subject and object roles reversed, achieving high attack success while evading a perplexity-based detector.

  2. Poison Once, Control Anywhere: Clean-Text Visual Backdoors in VLM-based Mobile Agents

    cs.CR 2025-06 conditional novelty 6.0 of 10

    Visual-only perturbations in fine-tuning screenshots can implant backdoors in VLM-based mobile agents, triggering attacker-chosen actions at inference.

  3. Robust Anti-Backdoor Instruction Tuning in LVLMs

    cs.CR 2025-06 conditional novelty 6.0 of 10

    A two-part defense, input diversity regularization plus anomalous activation sparsification, cuts backdoor attack success in adapter-tuned LVLMs to near zero in reported tests.

  4. AdInject: Real-World Black-Box Attacks on Web Agents via Advertising Delivery

    cs.CR 2025-05 conditional novelty 6.0 of 10

    Fake 'Close AD' ads make VLM web agents click them over 60% of the time, and near 100% in some settings.

  5. Backdoor Cleaning without External Guidance in MLLM Fine-tuning

    cs.CR 2025-05 conditional novelty 6.0 of 10

    BYE filters backdoored training images from MLLM fine-tuning by clustering low attention entropy across selected layers.

  6. BadVLA: Towards Backdoor Attacks on Vision-Language-Action Models via Objective-Decoupled Optimization

    cs.CR 2025-05 conditional novelty 6.0 of 10

    A two-stage, objective-decoupled training method embeds visual backdoors into VLA robot policies, achieving near-100% trigger-induced task failure with minimal clean-performance loss in simulation.

  7. Multimodal Fine-grained Reasoning for Post Quality Evaluation

    cs.LG 2025-07 reject novelty 5.0 of 10

    MFTRR combines local-global cross-modal attention, gating, and graph-based evidence reasoning to rank forum post quality, reporting NDCG@3 gains of up to 9.5 points over text-only baselines on new private datasets.

  8. Three Minds, One Legend: Jailbreak Large Reasoning Model with Adaptive Stacked Ciphers

    cs.CL 2025-05 conditional novelty 5.0 of 10

    A stacked-cipher jailbreak with adaptive code selection achieves 80.8% to 100% attack success on commercial large reasoning models.

Pith tools