8 papers
MMAligner: Safeguarding Multimodal Large Language Models through Representation Calibration
Shenyi Zhang, Keyan Guo, Zihao Wang +5
Multimodal large language models (MLLMs) often refuse unsafe text prompts yet generate harmful responses to semantically equivalent multimodal inputs. Existing defenses either rely…
VOID: Defeating Unauthorized Mimicry in Latent Diffusion Models
Chunlin Qiu, Ang Li, Tianxiao Huang +7
While Latent Diffusion Models (LDMs) have revolutionized visual synthesis, they are increasingly exploited for unauthorized mimicry of individuals. Existing defenses inject decepti…
Sparse Tokens Suffice: Jailbreaking Audio Language Models via Token-Aware Gradient Optimization
Zheng Fang, Xiaosen Wang, Shenyi Zhang +2
Jailbreak attacks on audio language models (ALMs) optimize audio perturbations to elicit unsafe generations, and they typically update the entire waveform densely throughout optimi…
Boosting Adversarial Transferability with Low-Cost Optimization via Maximin Expected Flatness
Chunlin Qiu, Ang Li, Yiheng Duan +4
Transfer-based attacks craft adversarial examples on white-box surrogate models and directly deploy them against black-box target models, offering model-agnostic and query-free thr…
Amulet: Fast TEE-Shielded Inference for On-Device Model Protection
Zikai Mao, Lingchen Zhao, Lei Xu +4
On-device machine learning (ML) introduces new security concerns about model privacy. Storing valuable trained ML models on user devices exposes them to potential extraction by adv…
Selective Masking Adversarial Attack on Automatic Speech Recognition Systems
Zheng Fang, Shenyi Zhang, Tao Wang +3
Extensive research has shown that Automatic Speech Recognition (ASR) systems are vulnerable to audio adversarial attacks. Current attacks mainly focus on single-source scenarios, i…