Research map / Image generation / Text-to-image diffusion
OrchJail: Jailbreaking Tool-Calling Text-to-Image Agents by Orchestration-Guided Fuzzing
Jianming Chen, Yawen Wang, Junjie Wang, Zhe Liu, Qing Wang, Fanjiang Xu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add orchestration abstraction and causal reasoning guidance (ablation) | OrchJail · GenArtist, CREA, LayerCraft | O-SR | Improved |
| add multi-objective LLM-judge scoring instead of random candidate selection (ablation) | OrchJail · GenArtist, CREA, LayerCraft | number of queries | Improved |
| add tool-aware seed generation (ablation) | OrchJail · GenArtist, CREA, LayerCraft | O-SR and queries | Improved |
| apply perplexity-based defense | OrchJail · GenArtist | O-SR: 72.63% -> 72.63% | No clear effect |
| apply SmoothLLM-S defense | OrchJail · GenArtist | O-SR: 72.63% -> 69.03% | Hurt |
| apply SmoothLLM-S defense | OrchJail · GenArtist | number of queries: 12.15 -> 16.74 | Hurt |
| orchestration-guided fuzzing vs JailFuzzer | OrchJail · GenArtist | O-SR: 63.48% -> 72.63% | Improved |
| orchestration-guided fuzzing vs JailFuzzer | OrchJail · LayerCraft | O-SR: 82.18% -> 91.68% | Improved |
| orchestration-guided fuzzing vs JailFuzzer | OrchJail · CREA | re-use success rate: 69.84% -> 79.11% | Improved |
| LLM-based mutation guided by orchestration vs token-level perturbation | OrchJail · GenArtist | prompt perplexity: 45.89 -> 35.91 (vs JailFuzzer) | Improved |
Closest papers by meaning
- SafePTR: Token-Level Jailbreak Defense in Multimodal LLMs via Prune-then-Restore Mechanism (NeurIPS 2025 · similarity 0.88)
- MT-JailBench: A Modular Benchmark for Understanding Multi-Turn Jailbreak Attacks (NeurIPS 2026 · similarity 0.88)
- Immune: Improving Safety Against Jailbreaks in Multi-modal LLMs via Inference-Time Alignment (CVPR 2025 · similarity 0.87)
- When the Prompt Becomes Visual: Vision-Centric Jailbreak Attacks for Large Image Editing Models (ICML 2026 · similarity 0.87)
- BlueSuffix: Reinforced Blue Teaming for Vision-Language Models Against Jailbreak Attacks (ICLR 2025 · similarity 0.87)
- h4rm3l: A Language for Composable Jailbreak Attack Synthesis (ICLR 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
