This article surveys efforts from the perspective of “trustworthy graph neural networks,” defined as those that incorporate core aspects such as robustness, explainability, privacy, fairness, accountability, and well-being.