Skip to content
conftrace_

preference hijacking

1 paper

Explore in graph

Co-occurring keywords · papers mentioning both

preference manipulation (1) image perturbation (1) inference-time attack (1) transferable perturbation (1) multimodal large language model (1) adversarial attack (1) model safety (1)

Papers

Phi: Preference Hijacking in Multi-modal Large Language Models at Inference Time EMNLP 2025