defense mechanisms 1information bottleneck 1intent-aware detection 1LLM agents 1memory injection attacks 1
From the 1 of 2 linked papers with an AI index.
2 papers
cs.AI2026
MIND: Lightweight and Effective Memory Injection Defense for LLM Agents via Intent-Aware Information Bottleneck
Dongyi Liu, Haixing He, Xiaobao Wu +1
The paper introduces MIND, a lightweight framework that uses an intent‑aware information bottleneck to detect and filter poisoned memory in large language model agents, reducing at…
cs.CR2026
Cross-Paradigm Graph Backdoor Attacks with Promptable Subgraph Triggers
Dongyi Liu, Jiangtong Li
Graph Neural Networks(GNNs) are vulnerable to backdoor attacks, where adversaries implant malicious triggers to manipulate model predictions. Existing trigger generators are often…