科研速览 · Science Skim继续刷下去 · Keep skimming →
◇ arXiv2026-08-12· cs.CL

When Explanations Betray Backdoors: Black-Box Auditing for Language Model Classifiers

Yang Liu, Ran Zou

原始摘要(英文原文)· Original abstract
Language model classifiers with explanations are used for moderation, routing, topic triage, and low-resource annotation. We study black-box auditing when the defender has only clean calibration data without trigger information but can ask the classifier for a label plus a short rationale or quoted evidence. We introduce Groundedness Drift, a lightweight score measuring whether the answer summary remains grounded in the input. Across two 7B backbones, five datasets, and four common non-adaptive OpenBackdoor-style attack families, Groundedness Drift achieves higher AUROC and lower residual target ASR than every compared detector in all cases at a nominal 5\% clean-FPR budget. We then evaluate Unsupported Groundedness, a multi-probe escalation for explanation-camouflage stress cases. Unsupported Groundedness improves signals but does not close the adaptive gap.
读原文 · Read the paper ↗

AI 追问PRO

登录后使用 AI 追问

讨论区

登录后参与讨论

相关论文 · Related

When Explanations Betray Backdoors: Black-Box Auditing for Language Model Classifiers — 科研速览 Science Skim