Researchers propose ROGUE, a framework for generating robust workflows for detecting audio deepfakes. ROGUE uses adversarial learning to select and execute detection tools, improving robustness to real-world perturbations and corruptions.