Abstract: Multilabel adversarial examples have become a threat to deep neural network models (DNNs). Most of the current work on multilabel adversarial examples are focused on white-box environments.