2 papers
cs.CR2026
Beyond the Prompt: Jailbreaking Function-Calling LLMs via Simulated Moderation Traces
Junlong Liu, Haobo Wang, Weiqi Luo +1
Jailbreak attacks remain a critical threat to the safe deployment of large language models (LLMs). While prior work has primarily studied attacks and defenses at the prompt level,…
cs.CV2026
Adversarial Attacks Against MLLMs via Progressive Resolution Processing and Adaptive Feature Alignment
Haobo Wang, Xiaorong Ma, Weiqi Luo +2
Adversarial perturbations can mislead Multimodal Large Language Models (MLLMs) recognize a benign image as a specific target object, posing serious risks in safety-critical scenari…