Adversarial images aligned with the latent distribution of unsafe content can force multimodal guard models to falsely reject safe user requests with up to 84% success.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
The Boy Who Cried Wolf: Adversarial Misclassification of Safe Inputs as Unsafe in Multimodal Guardrails
Adversarial images aligned with the latent distribution of unsafe content can force multimodal guard models to falsely reject safe user requests with up to 84% success.