Adversarial images optimized to map harmless text prefixes to toxic tokens jailbreak vision-language models more effectively than continuing toxic text.
Curriculum learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Benign-to-Toxic Jailbreaking: Inducing Harmful Responses from Harmless Prompts
Adversarial images optimized to map harmless text prefixes to toxic tokens jailbreak vision-language models more effectively than continuing toxic text.