Double-Blind Cleanser: Blindly Unlearning Backdoors Without Clean Data

Summary

Double-Blind Cleanser (DBC) is a backdoor unlearning framework for deep neural networks that needs neither a trusted clean dataset nor trigger recovery or poisoned-sample identification. It first uses catastrophic forgetting to remove backdoor behavior, then applies Sharpness-Aware Minimization (SAM) with importance sampling to restore legitimate functionality. The authors report that DBC neutralizes diverse backdoor attacks while maintaining model utility and placed first in the mitigation round of the TrojAI competition.

Related items

• HighCVE-2026-109331: Network-AI security control bypass in ClaudeHookBridge PreToolUse hookSimilar attack · NVD/CVE Database
• MediumCVE-2026-109296: Open WebUI missing authorization lets users reach disabled model backendsSimilar attack · NVD/CVE Database
• MediumCVE-2026-109292: Open WebUI information disclosure exposing session cookies on OAuth connectionsSimilar attack · NVD/CVE Database
• HighCVE-2026-108953: mac-shell-mcp OS command injection through execute_command argumentsSimilar attack · NVD/CVE Database
• HighCVE-2026-108836: Ollama server-side request forgery in Pull APISimilar attack · NVD/CVE Database