惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Microsoft Security Blog
Microsoft Security Blog
Jina AI
Jina AI
量子位
博客园 - 叶小钗
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
IT之家
IT之家
S
SegmentFault 最新的问题
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
小众软件
小众软件
Hugging Face - Blog
Hugging Face - Blog
雷峰网
雷峰网
博客园 - 聂微东
美团技术团队
Last Week in AI
Last Week in AI
罗磊的独立博客
酷 壳 – CoolShell
酷 壳 – CoolShell
博客园 - 三生石上(FineUI控件)
WordPress大学
WordPress大学
宝玉的分享
宝玉的分享
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
博客园_首页
V
Visual Studio Blog
大猫的无限游戏
大猫的无限游戏
The Cloudflare Blog

Paper Index on ACL Anthology

A Bounded Coordination-Support Capability for Multi-Party Settings: Task-State Monitoring in Firefighter Incident Command A Dataset of Latin Etymologies Extracted from Wiktionary An Efficient Approach for Answering Not Readily Attainable Questions for RAG-based Applications Automated German Alt Text Generation for News Charts Call Support Copilot: A Reproducible Multimodal System for Speech Emotion Recognition, Intent Understanding, and Agent Assistance Can Large Language Models Replace Statistical Software? Code-Switching Detection in Multilingual Child Speech with SwissBERT Concept Extraction and Webb’s Depth of Knowledge: Comparing LLM Question Generation Pipelines for Educational Assessment Data Augmentation for Historical NER: A Systematic Comparison of Lexical and LLM-based Approaches Enhancing Retrieval via Cognitively Motivated Document Expansion Extending the Contact Hypothesis: Cross-Linguistic Evaluation of Religion and Nationality Bias When Prompting LLMs in German and Icelandic Extracting Article-Level Legal Dependencies from Swiss Federal Law using LLMs How Good is AI on Swiss Voting Booklets? A Multilingual OCR and Alignment Benchmark Optimizing Large Language Models for Robust Domain-Specific Text-to-SQL: From Prompting to Preference Alignment Proceedings of the 11th Edition of the Swiss Text Analytics Conference Reinforcement Learning for Latent-Space Thinking in LLMs RUMLEM: A Dictionary-Based Lemmatizer for Romansh Skill Extraction from Resumes and Job Offers across Six Languages Text vs. Phoneme Intermediates for Low-Resource Swiss German The Same Email, Signed Differently: Testing Negotiation Bias and Recommendation Stability in LLMs Which Skills Debate Reaches the Public? Comparing Scientific Literature and Media Coverage of AI and LLM Skill Impacts (2022–2025) Controlling Language and Style of Multi-lingual Generative Language Models with Control Vectors Hybrid Human-LLM Corpus Construction and LLM Evaluation for the Caused-Motion Construction Implicit and Indirect: Detecting Face-threatening and Paired Actions in Asynchronous Online Conversations Northern European Journal of Language Technology, Volume 11 A modular architecture for creating multimodal embodied agents with an episodic Knowledge Graph as an explainable and controllable long-term memory A Neural Approach to Discourse Relation Signal Detection An Analysis of Japanese Sentence-final Particle Yone: Compare Yone and Ne in Response Attribution and the discourse structure of reports Automatic Detection of the Bulgarian Evidential Renarrative
aclanthology.org
2026-06-22 · via Paper Index on ACL Anthology

Abstract

The rapid proliferation of Multimodal Large Language Models (MLLMs) has ushered in the era of the “Agentic Economy,” where Mobile Agents autonomously execute high-stakes financial transactions. While these agents demonstrate impressive operational capabilities, their adversarial robustness remains a glaring blind spot. In this paper, we identify a systemic vulnerability termed Visual Dominance Hallucination (VDH), where imperceptible adversarial visual cues can act as a “super-stimulus,” overriding textual price evidence in our evaluated screenshot-based price-constrained settings and forcing the agent into irrational economic decisions. We propose PriceBlind, a stealthy, white-box adversarial attack framework for controlled screenshot-based evaluation. Unlike prior works that rely on conspicuous artifacts like pop-ups, PriceBlind exploits the modality gap in CLIP-based encoders via a novel Semantic-Decoupling Loss. Rather than literally making a luxury item “look cheap,” this regularizer weakens the consistency between high-price text and visual value cues by aligning the image embedding with a low-cost/value-associated anchor region while preserving pixel-level fidelity. On our main E-ShopBench benchmark with clear price constraints, screenshot-based white-box evaluation yields ASRs around 80% on the evaluated agents. Under the evaluated single-turn coordinate-selection protocol in a simplified layout-aware setting, our Ensemble-DI-FGSM strategy also yields non-trivial black-box transfer, with ASR roughly 35–41% across GPT-4o, Gemini-1.5-Pro, and Claude-3.5-Sonnet. In the same screenshot-based setting, standard robust encoders reduce ASR only partially, while a Verify-then-Act stack with robust encoders lowers ASR to below 10% at some clean-accuracy cost.

Anthology ID:
2026.findings-acl.788
Volume:
Findings of the Association for Computational Linguistics: ACL 2026
Month:
July
Year:
2026
Address:
San Diego, California, United States
Editors:
Maria Liakata, Viviane P. Moreira, Jiajun Zhang, David Jurgens
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
16059–16073
Language:
URL:
https://aclanthology.org/2026.findings-acl.788/
DOI:
Bibkey:
Cite (ACL):
Jiachen Qian and Zhaolu Kang. 2026. "Penny Wise, Pixel Foolish": Bypassing Price Constraints in Multimodal Agents via Visual Adversarial Perturbations. In Findings of the Association for Computational Linguistics: ACL 2026, pages 16059–16073, San Diego, California, United States. Association for Computational Linguistics.
Cite (Informal):
“Penny Wise, Pixel Foolish”: Bypassing Price Constraints in Multimodal Agents via Visual Adversarial Perturbations (Qian & Kang, Findings 2026)
Copy Citation:
PDF:
https://aclanthology.org/2026.findings-acl.788.pdf
Checklist:
 2026.findings-acl.788.checklist.pdf