Graph neural networks (GNNs) have been widely applied in a variety of domains. However, the very ability of graphs to represent complex data structures is both the key strength of GNNs and a major source of their vulnerability. Recent studies have shown that attacking GNNs by maliciously perturbing the underlying graph can severely degrade their performance. For attack methods, the central challenge is to maintain attack effectiveness while remaining difficult to detect. Most existing attacks require modifying the graph structure, such as adding or deleting edges, which is relatively easy to notice. To address this problem, this paper proposes a new attack model that employs spectral adversarial robustness evaluation to quantitatively analyze the vulnerability of each edge in a graph. By precisely targeting the weakest links, our method can achieve effective attacks without changing the connectivity pattern of edges in the graph, for example by subtly adjusting the weights of a small subset of the most vulnerable edges. We apply the proposed method to attack several classical graph neural network architectures, and experimental results show that our attack is highly effective.

Graph Neural Networks (GNNs) have emerged as a dominant paradigm for learning on graph-structured data, thanks to their ability to jointly exploit node features and relational information encoded in the graph topology. This joint modeling, however, also introduces a critical weakness: perturbations or noise in either the structure or the features can be amplified through message passing, making GNNs highly vulnerable to adversarial attacks and spurious connections. In this work, we introduce a pruning framework that leverages adversarial robustness evaluation to explicitly identify and remove fragile or detrimental components of the graph. By using robustness scores as guidance, our method selectively prunes edges that are most likely to degrade model reliability, thereby yielding cleaner and more resilient graph representations. We instantiate this framework on three representative GNN architectures and conduct extensive experiments on benchmarks. The experimental results show that our approach can significantly enhance the defense capability of GNNs in the high-perturbation regime.