Bergeron introduces a framework to enhance the robustness of AI models against adversarial attacks without additional training.