BadSKP: Backdoor Attacks on Knowledge Graph-Enhanced LLMs with Soft Prompts

arXiv:2605.11996v1 Announce Type: new Abstract: Recent knowledge graph (KG)-enhanced large language models (LLMs) move beyond purely textual knowledge augmentation by encoding retrieved subgraphs into continuous soft prompts via graph neural networks, introducing a graph-conditioned channel that operates alongside the standard text interface. However, existing backdoor attacks are largely designed for the textual channel, and their effectiveness against this dual-channel architecture remains unc