AI Paper Atlas

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

CVPR 2026 · Adversarial robustness and security › Backdoor and poisoning · approach: Detecting poisoned samples · 2 citations (OpenAlex)

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 changedWhereMeasuredEffect
add MirrorVerify cross-prompt stability check (ablation)BlackMirror · Avg over 8 attacksFPR: 93.06 -> 15.09Improved
replace UFID image-level similarity with pattern-level deviation detectionblack-box T2I backdoor detection · ObjRepAtt BadT2IF1: 66.67 -> 86.96Improved
replace UFID image-level similarity with pattern-level deviation detectionblack-box T2I backdoor detection · ObjRepAtt EvilEditF1: 60.87 -> 85.71Improved
use BlackMirror instead of UFID on fixed-image attackblack-box T2I backdoor detection · FixImgAtt VillanDiffusionF1: 90.91 -> 80.00Hurt
add majority voting over K VLM object extractions (ablation)BlackMirror MirrorMatch · ObjRepAtt attacksFPR: average drop of around 5%Improved
add majority voting over K VLM object extractions (ablation)BlackMirror MirrorMatch · ObjRepAtt attackstime per sample: decreases by approximately 4 secondsImproved
increase number of generations N in MirrorVerify (ablation)BlackMirror · Avg over attacksFPRImproved
raise decision threshold tau up to 0.999 (ablation)BlackMirror · Avg over attacksFPR / F1Improved
raise decision threshold tau too high (ablation)BlackMirror · Avg over attacksrecallHurt
use global CLIP instruction-response similarity as detectornaive baseline · BadT2I, EvilEdit, PaaS ObjRepAtt etc.F1: 65.55 avg vs 89.46 for BlackMirrorHurt
run object, patch and style branches in parallelBlackMirror · all attackscompute costNo 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.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy