BlackMirror：基于指令-响应偏差的文本到图像模型黑盒后门检测
BlackMirror: Black-Box Backdoor Detection for Text-to-Image Models via Instruction-Response Deviation
提出一种免训练、可插拔的黑盒后门检测思路：不依赖图像级相似度，而是把生成图像中的局部语义模式与指令对齐，并在多提示下检验偏差稳定性，可直接用于 MaaS 场景下对第三方文生图模型的审计。
原文依据 only partial semantic patterns within the generated image are steadily manipulated, while the rest of the content remains diverse or benign
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论本文研究在黑盒设定下检测被植入后门的文本到图像模型这一具有挑战性的任务，并提出一种新的检测框架 BlackMirror。现有方法通常依赖分析图像级相似度，其假设是后门触发生成的结果在不同样本之间表现出强一致性。然而，这些方法难以泛化到近期出现的后门攻击，因为后门生成结果在视觉上可能呈现多样性。BlackMirror 的动机来自一个观察：在各类后门攻击中，生成图像内只有部分语义模式被稳定地操纵，其余内容仍保持多样或良性。据此，BlackMirror 由两个组件构成：MirrorMatch 将视觉模式与对应指令对齐以检测语义偏差；MirrorVerify 评估这些偏差在不同提示下的稳定性，从而区分真实后门行为与良性响应。BlackMirror 是一个通用的免训练框架，可作为即插即用模块部署于模型即服务（MaaS）应用中。大量实验表明，BlackMirror 在多种攻击下均能实现准确检测。代码见 https://github.com/Ferry-Li/BlackMirror。
英文原文摘要 1321 字符
This paper investigates the challenging task of detecting backdoored text-to-image models under black-box settings and introduces a novel detection framework BlackMirror. Existing approaches typically rely on analyzing image-level similarity, under the assumption that backdoor-triggered generations exhibit strong consistency across samples. However, they struggle to generalize to recently emerging backdoor attacks, where backdoored generations can appear visually diverse. BlackMirror is motivated by an observation: across backdoor attacks, only partial semantic patterns within the generated image are steadily manipulated, while the rest of the content remains diverse or benign. Accordingly, BlackMirror consists of two components: MirrorMatch, which aligns visual patterns with the corresponding instructions to detect semantic deviations; and MirrorVerify, which evaluates the stability of these deviations across varied prompts to distinguish true backdoor behavior from benign responses. BlackMirror is a general, training-free framework that can be deployed as a plug-and-play module in Model-as-a-Service (MaaS) applications. Comprehensive experiments demonstrate that BlackMirror achieves accurate detection across a wide range of attacks. Code is available at this https://github.com/Ferry-Li/BlackMirror.