Multi-turn image generators complete hateful visual stories in over 80% of attempts, existing per-image moderators miss most of them, and interaction-aware monitoring detects 92.6-97.3% of such stories.
Jail- breakRadar: Comprehensive Assessment of Jailbreak Attacks Against LLMs
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Innocent Panels, Hateful Stories: Evaluating and Detecting Hateful Intent in Multi-Turn Visual Story Generation
Multi-turn image generators complete hateful visual stories in over 80% of attempts, existing per-image moderators miss most of them, and interaction-aware monitoring detects 92.6-97.3% of such stories.