Research map / Adversarial robustness and security / Backdoor and poisoning
BlackMirror: Black-Box Backdoor Detection for Text-to-Image Models via Instruction-Response Deviation
Feiran Li, Qianqian Xu, Shilong Bao, Zhiyong Yang, Xilin Zhao, Xiaochun Cao, Qingming Huang
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 MirrorVerify cross-prompt stability check (ablation) | BlackMirror · Avg over 8 attacks | FPR: 93.06 -> 15.09 | Improved |
| replace UFID image-level similarity with pattern-level deviation detection | black-box T2I backdoor detection · ObjRepAtt BadT2I | F1: 66.67 -> 86.96 | Improved |
| replace UFID image-level similarity with pattern-level deviation detection | black-box T2I backdoor detection · ObjRepAtt EvilEdit | F1: 60.87 -> 85.71 | Improved |
| use BlackMirror instead of UFID on fixed-image attack | black-box T2I backdoor detection · FixImgAtt VillanDiffusion | F1: 90.91 -> 80.00 | Hurt |
| add majority voting over K VLM object extractions (ablation) | BlackMirror MirrorMatch · ObjRepAtt attacks | FPR: average drop of around 5% | Improved |
| add majority voting over K VLM object extractions (ablation) | BlackMirror MirrorMatch · ObjRepAtt attacks | time per sample: decreases by approximately 4 seconds | Improved |
| increase number of generations N in MirrorVerify (ablation) | BlackMirror · Avg over attacks | FPR | Improved |
| raise decision threshold tau up to 0.999 (ablation) | BlackMirror · Avg over attacks | FPR / F1 | Improved |
| raise decision threshold tau too high (ablation) | BlackMirror · Avg over attacks | recall | Hurt |
| use global CLIP instruction-response similarity as detector | naive baseline · BadT2I, EvilEdit, PaaS ObjRepAtt etc. | F1: 65.55 avg vs 89.46 for BlackMirror | Hurt |
| run object, patch and style branches in parallel | BlackMirror · all attacks | compute cost | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- UFID — UFID: A Unified Framework for Black-box Input-level Backdoor Detection on Diffusion Models (AAAI 2025) · better on 8 of 9 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Test-Time Multimodal Backdoor Detection by Contrastive Prompting (ICML 2025 · similarity 0.86)
- Revisiting Backdoor Attacks against Large Vision-Language Models from Domain Shift (CVPR 2025 · similarity 0.86)
- UFID: A Unified Framework for Black-box Input-level Backdoor Detection on Diffusion Models (AAAI 2025 · similarity 0.86)
- Detecting Backdoor Samples in Contrastive Language Image Pretraining (ICLR 2025 · similarity 0.85)
- Backdooring Vision-Language Models with Out-Of-Distribution Data (ICLR 2025 · similarity 0.85)
- Semantic Shield: Defending Vision-Language Models Against Backdooring and Poisoning via Fine-grained Knowledge Alignment (CVPR 2024 · similarity 0.84)
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.
