Recent advances in generative AI have significantly expanded into the realms of art and music. This development has opened up a vast realm of possibilities, pushing the boundaries of human creativity into unexplored frontiers. However, as generative AI advances, it can replicate artistic styles and produce new artwork, posing significant concerns for the perceived rarity and value of artists' creations. In response to these challenges, it is becoming increasingly crucial to establish and enforce protective measures that safeguard artists' copyrighted work from unauthorized exploitation by generative AI models. In this paper, we introduce the first defensive mechanism, HARMONYCLOAK, to prevent the exploitative use of artwork, specifically in the context of instrumental music, by generative AI models. Particularly, HARMONYCLOAK employs imperceptible error-minimizing noise to make the model's generative loss approach zero for these perturbed music data, tricking the model into believing nothing can be learned so as to disrupt their attempts to replicate musical structures and styles. By using a set of intra-track and inter-track objective metrics and a subjective user study, extensive experiments on three state-of-the-art music generative AI models (i.e., MuseGAN, SymphonyNet, and MusicLM) validate the effectiveness and applicability of Harmonycloak1.1.Audio examples of the unlearnable music examples are available for listening at https://mosis.eecs.utk.edu/harmonycloak.html. in both white-box and black-box settings.

Federated learning (FL) provides a variety of privacy advantages by allowing clients to collaboratively train a model without sharing their private data. However, recent studies have shown that private information can still be leaked through shared gradients. To further minimize the risk of privacy leakage, existing defenses usually require clients to locally modify their gradients (e.g., differential privacy) prior to sharing with the server. While these approaches are effective in certain cases, they regard the entire data as a single entity to protect, which usually comes at a large cost in model utility. In this paper, we seek to reconcile utility and privacy in FL by proposing a user-configurable privacy defense, RecUP-FL, that can better focus on the user-specified sensitive attributes while obtaining significant improvements in utility over traditional defenses. Moreover, we observe that existing inference attacks often rely on a machine learning model to extract the private information (e.g., attributes). We thus formulate such a privacy defense as an adversarial learning problem, where RecUP-FL generates slight perturbations that can be added to the gradients before sharing to fool adversary models. To improve the transferability to un-queryable black-box adversary models, inspired by the idea of meta-learning, RecUP-FL forms a model zoo containing a set of substitute models and iteratively alternates between simulations of the white-box and the black-box adversarial attack scenarios to generate perturbations. Extensive experiments on four datasets under various adversarial settings (both attribute inference attack and data reconstruction attack) show that RecUP-FL can meet user-specified privacy constraints over the sensitive attributes while significantly improving the model utility compared with state-of-the-art privacy defenses.