Are LLM-Enhanced GNNs Privacy-Safe?

Abstract

Large language models (LLMs) have recently advanced graph neural networks (GNNs) by enriching node representations with semantic information, giving rise to LLM-enhanced GNNs that achieve substantial performance gains. However, their vulnerability to privacy attacks, in which adversaries infer sensitive information from model outputs, remains largely underexplored. To bridge this gap, we present a systematic evaluation of privacy risks in LLM-enhanced GNNs through a unified framework consisting

Categories

Framework mappings

Suggested from the entry's categories.

@misc{he2026are,
  title = {{Are LLM-Enhanced GNNs Privacy-Safe?}},
  author = {Long-Zhu He and Zekun Wen and Chaozhuo Li and Sen Su},
  year = {2026},
  month = aug,
  eprint = {2608.25727},
  archivePrefix = {arXiv},
  url = {https://www.semanticscholar.org/paper/9f818c0b7d9af6900dfa284aeb0cbb0b244e1fb0}
}