Images are Achilles' Heel of Alignment: Exploiting Visual Vulnerabilities for Jailbreaking Multimodal Large Language Models.
Yifan LiHangyu GuoKun ZhouWayne Xin ZhaoJi-Rong WenPublished in: CoRR (2024)
Keyphrases
- language model
- n gram
- image data
- language modeling
- visual data
- probabilistic model
- image database
- image retrieval
- information retrieval
- document retrieval
- image collections
- language modelling
- image features
- image classification
- smoothing methods
- context sensitive
- similarity measure
- multimedia
- image annotation
- language model for information retrieval
- retrieval model
- test collection
- visual features
- bag of words
- image regions
- speech recognition
- relevance model
- query expansion
- web search
- statistical language models
- low level