analiza aiqbench
This research introduces UnivIntruder, an adversarial attack framework leveraging the CLIP model to generate highly effective perturbations against diverse deep neural networks without needing access to target training data. The study demonstrates significant vulnerabilities in real-world systems, including image search engines and large vision-language models like GPT-4, achieving over 80% success rates even when direct queries are restricted. These findings highlight critical security gaps that necessitate a reevaluation of robustness paradigms for AI applications operating under constrained query environments.
One Surrogate to Fool Them All: Universal, Transferable, and Targeted Adversarial Attacks with CLIP
arXiv:2505.19840v3 Announce Type: replace-cross Abstract: Deep Neural Networks (DNNs) have achieved widespread success yet remain prone to adversarial attacks. Typically, such attacks either involve frequent queries to the target model or rely on surrogate models closely mirroring the target model -- often trained with subsets of the target model's training data -- to achieve high attack success rates through transferability. However, in realistic scenarios where training data is inaccessible and excessive queries can raise alarms, crafting adversarial examples becomes more challenging. In this paper, we present UnivIntruder, a novel attack framework that relies solely on a single, publicly available CLIP model and publicly available datasets. By using textual concepts, UnivIntruder generates universal, transferable, and targeted adversarial perturbations that mislead DNNs into misclassifying inputs into adversary-specified classes defined by textual concepts. Our extensive experiments show that our approach achieves an Attack Success Rate (ASR) of up to 85% on ImageNet and over 99% on CIFAR-10, significantly outperforming existing transfer-based methods. Additionally, we reveal real-world vulnerabilities, showing that even without querying target models, UnivIntruder compromises image search engines like Google and Baidu with ASR rates up to 84%, and vision language models like GPT-4 and Claude-3.5 with ASR rates up to 80%. These findings underscore the practicality of our attack in scenarios where traditional avenues are blocked, highlighting the need to reevaluate security paradigms in AI applications.
Abstract:Deep Neural Networks (DNNs) have achieved widespread success yet remain prone to adversarial attacks. Typically, such attacks either involve frequent queries to the target model or rely on surrogate models closely mirroring the target model -- often trained with subsets of the target model's training data -- to achieve high attack success rates through transferability. However, in realistic scenarios where training data is inaccessible and excessive queries can raise alarms, crafting adversarial examples becomes more challenging. In this paper, we present UnivIntruder, a novel attack framework that relies solely on a single, publicly available CLIP model and publicly available datasets. By using textual concepts, UnivIntruder generates universal, transferable, and targeted adversarial perturbations that mislead DNNs into misclassifying inputs into adversary-specified classes defined by textual concepts. Our extensive experiments show that our approach achieves an Attack Success Rate (ASR) of up to 85% on ImageNet and over 99% on CIFAR-10, significantly outperforming existing transfer-based methods. Additionally, we reveal real-world vulnerabilities, showing that even without querying target models, UnivIntruder compromises image search engines like Google and Baidu with ASR rates up to 84%, and vision language models like GPT-4 and Claude-3.5 with ASR rates up to 80%. These findings underscore the practicality of our attack in scenarios where traditional avenues are blocked, highlighting the need to reevaluate security paradigms in AI applications.
22 pages, 15 figures, 18 tables. To appear in the Proceedings of The ACM Conference on Computer and Communications Security (CCS), 2025
Cryptography and Security (cs.CR); Machine Learning (cs.LG)
arXiv:2505.19840 [cs.CR]
(or arXiv:2505.19840v3 [cs.CR] for this version)
https://doi.org/10.48550/arXiv.2505.19840 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 338 cuvinte
Subiecte conexe
One Attack to Fool Them All: Highly Transferable Black-Box Adversarial Attacks on Frontier MLLMs
arXiv · 30 sept. 2026
Information-Geometric Inverse Distillation for Enhancing Adversarial Transferability
arXiv · 18 sept. 2026
Improving Adversarial Transferability on Vision-Language Pre-training Models via Surrogate-Specific Bias Correction
arXiv · 10 iun. 2026
What Makes Adversarial Examples Transfer Across Deepfake Detectors?
arXiv · 11 sept. 2026