LATIDIA · Ciberseguridad

Contención de puerta trasera a través de cuarentena experta y apagado en LLM

arXiv: 2610.00663v2Tipo de anuncio: replace-cross Resumen: Los modelos de lenguaje grande (LLM) con puerta trasera pueden comportarse normalmente en entradas benignas mientras producen salidas especificadas por el atacante bajo disparadores ocultos. Defensas existentes

La noticia

arXiv:2610.00663v2 Announce Type: replace-cross Abstract: Backdoored large language models (LLMs) can behave normally on benign inputs while producing attacker-specified outputs under hidden triggers. Existing defenses span four stages---prior-training, in-training, post-training, and inference-time---and share one of two underlying strategies: either suppress backdoor learning (by filtering poisoned data or interrupting its acquisition during optimization) or learn, then purify (by repairing model weights or gating inputs after a fully backdoored model has formed). We propose a third strategy, learn, but channel: allow backdoor formation during training but route it into a designated, quarantined component that can be disabled at deployment. To this end, we propose Quarantined Expert Shutdown QES, a computationally efficient containment